Resting state fMRI study of brain activation using rTMS in ... · We used independent component...

Resting state fMRI study of brain activation

using rTMS in rats

Bhedita Jaya Seewoo1, BSc

(Student no: 31925952)

Supervisors:

Dr Sarah Etherington1, BSc, PhD

A/Prof Jennifer Rodger2, BSc, PhD

Dr Kirk Feindel3, BSc, PhD

SUBMITTED ON: 31ST OCTOBER 2016

Bachelor of Science Honours in Biomedical Science

1 School of Veterinary and Life Sciences, Murdoch University 2 Experimental and Regenerative Neuroscience (EaRN), School of Animal Biology, University of Western Australia 3Centre for Microscopy, Characterisation and Analysis (CMCA), University of Western Australia

DECLARATION

I declare this thesis is my own account of my research and contains as its main content

work, which has not been previously submitted for a degree at any tertiary education

institution.

Bhedita Jaya Seewoo

Resting state fMRI study of brain activation using rTMS in rats Bhedita J. Seewoo | 1

ABSTRACT

Background and purpose: Repetitive transcranial magnetic stimulation (rTMS) is a non-

invasive neuromodulation technique used to treat many neurological and psychiatric

conditions. However, not much is known about the mechanisms underlying its efficacy

because human rTMS studies are mostly non-invasive while most animal studies are

invasive. Invasive animal studies allow for cellular and molecular changes to be detected

and hence, have been able to show that rTMS may alter synaptic plasticity in the form of

long-term potentiation. This is the first rodent study using non-invasive resting state

functional magnetic resonance imaging (rs-fMRI) to examine the effects of low-intensity

rTMS (LI-rTMS) in order to provide a more direct comparison to human studies.

Methods: rs-fMRI data were acquired before and after 10 minutes of LI-rTMS intervention

at one of four frequencies—1 Hz, 10 Hz, biomimetic high frequency stimulation (BHFS) and

continuous theta burst stimulation (cTBS)—in addition to sham. We used independent

component analysis to uncover changes in the default mode network (DMN) induced by

each rTMS protocol.

Results: There were considerable rTMS-related changes in the DMN. Specifically, (1) the

synchrony of resting activity of the somatosensory cortex was decreased ipsilaterally

following 10 Hz stimulation, increased ipsilaterally following cTBS, and decreased bilaterally

following 1 Hz stimulation and BHFS; (2) the motor cortex showed bilateral changes

following 1 Hz and 10 Hz stimulation, an ipsilateral increase in synchrony of resting activity

following cTBS, and a contralateral decrease following BHFS; and (3) in the hippocampus,

10 Hz stimulation caused an ipsilateral decrease while 1 Hz and BHFS caused a bilateral

decrease in synchrony. There was no change in the correlation of the hippocampus induced

by cTBS.

Conclusion: The present findings suggest that LI-rTMS can modulate functional links within

the DMN of rats. LI-rTMS can induce changes in the cortex, as well as in remote brain regions

such as the hippocampus when applied to anaesthetised rats and the pattern of these

changes depends on the frequency used, with 10 Hz stimulation, BHFS and cTBS causing

mostly ipsilateral changes in synchrony of activity in the DMN and 1 Hz stimulation causing


bilateral changes in synchrony, with the contralateral changes being more prominent than

ipsilateral changes. Hence, combined rTMS-fMRI emerges as a powerful tool to visualise

rTMS-induced cortical connectivity changes at a high spatio-temporal resolution and help

unravel the physiological processes underlying these changes in the cortex and

interconnected brain regions.

Keywords- rTMS | resting-state | functional MRI | ICA | default mode network


TABLE OF CONTENTS

ABSTRACT ...................................................................................................................... 1

TABLE OF CONTENTS ...................................................................................................... 3

LIST OF FIGURES AND TABLES ......................................................................................... 6

LIST OF ABBREVIATIONS................................................................................................. 8

ACKNOWLEDGEMENTS .................................................................................................. 9

CHAPTER ONE: INTRODUCTION .................................................................................... 10

1.1 OVERVIEW ............................................................................................................................................... 10

1.2 TRANSCRANIAL MAGNETIC STIMULATION (TMS) ................................................................................... 12

1.2.1 The principle behind TMS ................................................................................................................ 12

1.2.2 Different rTMS paradigms ............................................................................................................... 14

1.2.2.1 The different rTMS frequencies and patterns of stimulation ................................................................... 14

1.2.2.2 The different rTMS intensities .................................................................................................................. 18

1.2.3 The after-effects of rTMS – Synaptic Plasticity (LTP and LTD) ......................................................... 18

1.2.4 Variability in rTMS effects in humans ............................................................................................. 21

1.2.5 rTMS delivery in humans vs. rodents............................................................................................... 23

1.3 MEASURING THE EFFECTS OF RTMS NON-INVASIVELY ............................................................................ 25

1.3.1 Detecting the effects of rTMS using EEG and EMGs........................................................................ 25

1.3.2 Detecting rTMS-induced changes using brain imaging techniques ................................................ 26

1.3.3 fMRI study of brain activation ......................................................................................................... 26

1.3.3.1 Background to MRI ................................................................................................................................... 26

1.3.3.2 Use of fMRI to determine changes induced by rTMS ............................................................................... 29

1.3.3.3 Resting-state fMRI, the default mode network and rTMS ....................................................................... 31

1.4 MEASURING THE EFFECTS OF RTMS IN ANIMALS (INVASIVE) .................................................................. 33

1.4.1 Induction of the expression of plasticity-related genes ................................................................... 33

1.4.2 Up-regulation of growth factor expression ..................................................................................... 35

1.4.3 Changes in receptor expression....................................................................................................... 36

1.5 PRESENT STUDY ....................................................................................................................................... 37

1.5.1 Hypotheses and Aims ...................................................................................................................... 37

CHAPTER TWO: MATERIALS AND METHODS ................................................................. 38

2.1 ANIMALS ................................................................................................................................................. 38


2.2 OVERVIEW OF THE EXPERIMENTAL DESIGN ............................................................................................ 38

2.3 ISOFLURANE ANAESTHESIA PROCEDURE AND MONITORING ................................................................. 39

2.4 LOW-INTENSITY REPETITIVE TRANSCRANIAL MAGNETIC STIMULATION ................................................ 40

2.4.1 Coil Structure and Positioning ......................................................................................................... 40

2.4.2 Stimulation Protocols ...................................................................................................................... 41

2.5 MRI SCAN PROCEDURE ............................................................................................................................ 42

2.5.1 Equipment ....................................................................................................................................... 42

2.5.2 Image acquisition parameters ........................................................................................................ 42

2.6 DATA ANALYSIS ....................................................................................................................................... 44

2.6.1 fMRI data processing ...................................................................................................................... 44

2.6.2 fMRI statistical analyses.................................................................................................................. 46

CHAPTER THREE: RESULTS ............................................................................................ 49

3.1 PHYSIOLOGICAL RECORDING .................................................................................................................. 49

3.2 GROUP-ICA COMPONENTS ..................................................................................................................... 50

3.2.1 Default mode network in the pre-stimulation group ...................................................................... 50

3.2.2 Reproducibility over time and between subjects in the pre-stimulation group .............................. 53

3.2.3 Pre- vs. post-stimulation resting-state activity ............................................................................... 54

3.2.3.1 Dual regression results ............................................................................................................................. 54

3.2.3.2 Comparison of DMN in ICA components .................................................................................................. 54

3.2.4 Effect of smoothing ......................................................................................................................... 58

CHAPTER FOUR: DISCUSSION ....................................................................................... 61

4.1 THE DEFAULT MODE NETWORK BEFORE STIMULATION ......................................................................... 61

4.2 PRE- VS. POST-STIMULATION RESTING-STATE ACTIVITY ......................................................................... 62

4.2.1 Sham stimulation does not affect the DMN .................................................................................... 63

4.2.2 Active LI-rTMS disrupts the DMN in rat brain ................................................................................. 63

4.2.2.1 Somatosensory and Motor cortex ............................................................................................................ 64

4.2.2.2 Hippocampus ........................................................................................................................................... 66

4.3 DUAL REGRESSION: PRE- VS. POST- ACTIVE RTMS STIMULATION ........................................................... 67

4.4 ASSESSMENT OF ANALYSIS PARAMETERS ............................................................................................... 68

4.4.1 Effect of smoothing ......................................................................................................................... 68

4.4.2 Possible issues with rs-fMRI and rTMS in rodents ........................................................................... 70

4.4.2.1 Maintaining stable experimental conditions ............................................................................................ 70

4.4.2.2 Use of anaesthetics .................................................................................................................................. 71

4.5 FUTURE DIRECTIONS ............................................................................................................................... 72

4.6 CONCLUSION ........................................................................................................................................... 73


REFERENCES................................................................................................................. 74

APPENDICES................................................................................................................. 93

APPENDIX 1: FMRI IMAGE PRE-PROCESSING PROTOCOL .............................................................................. 93

APPENDIX 2: DATA ANALYSIS ........................................................................................................................ 98

APPENDIX 3: BRAIN ATLAS .......................................................................................................................... 103


LIST OF FIGURES AND TABLES

Figure 1. Neuronal activation by TMS using "figure-of-eight" coil. ..................................... 12

Figure 2. Transcranial magnetic stimulation (TMS) applied over the motor cortex activates neurons leading to a corticospinal volley down the corticospinal tract. ............................. 13

Figure 3. Repetitive transcranial magnetic stimulation (rTMS) protocols. .......................... 15

Figure 4. The size of motor-evoked potentials (MEPs) is smaller after continuous theta burst stimulation (cTBS) and more pronounced after intermittent theta burst stimulation (iTBS) compared to baseline. .......................................................................................................... 17

Figure 5. Intracellular signalling mechanisms for long-term potentiation (LTP) and long-term depression (LTD). .................................................................................................................. 20

Figure 6. Relative magnitude (y-axis) and potential significance (x-axis) of determinants of plasticity induction by non-invasive brain stimulation in persistently determining the direction and magnitude of the induced plasticity. ............................................................. 23

Figure 7. Summary of available and desired coils for rTMS in animal studies that are equivalent to those used in human studies. ........................................................................ 24

Figure 8. Anatomical MRI image of a healthy human brain showing the left (1) and right (2) dorsolateral prefrontal cortex and left anterior cingulate cortex (3) in trans-axial orientation. .............................................................................................................................................. 27

Figure 9. The difference in blood oxygenation at resting state (a) and when the brain is active (b). ........................................................................................................................................ 28

Figure 10. Timeline of an experiment for one rat and protocol for each session. .............. 39

Figure 11. Animal position on the imaging cradle during LI-rTMS stimulation and fMRI scans. .............................................................................................................................................. 40

Figure 12. Stimulation coil positioning in relation to rat’s head and brain. ........................ 41

Figure 13. Representative physiological recordings for two animals demonstrating that body temperature (degrees Fahrenheit, A), respiratory rate (breaths per minute, B), oxygen saturation (%, C) and pulse (beats per minute, D) were stable. .......................................... 49

Figure 14. Independent component maps of pre-stimulation rs-fMRI group overlaid on 3D-rendered standard Sprague Dawley brain atlas. .................................................................. 51

Figure 15. Thresholded independent component maps showing the default mode network in the pre-stimulation rs-fMRI group-ICA dataset. .............................................................. 52

Figure 16. Two independent components — one axial slice (A) and one coronal slice (B) — from group-ICA classified as ‘noise’ components. ............................................................... 53

Figure 17. Reproducibility between sessions and between subjects of a representative pre-stimulation group-ICA component. ...................................................................................... 54

Figure 18. Homologous ICA components showing that there is no change in the default mode network in isoflurane-anaesthetized rats before and after sham stimulation ......... 55


Figure 19. Homologous group-ICA components showing changes in the default mode network of isoflurane-anaesthetized rats before and after rTMS stimulation at four frequencies: 10 Hz, BHFS, 1 Hz and cTBS. ............................................................................ 56

Figure 20. Effect of smoothing on individual ICA components for the pre-stimulation group (A) and the t-stat activation maps of post- minus pre- 10 Hz stimulation of a representative animal (B).............................................................................................................................. 60

Supplementary Figure 1. Characteristic power spectra and time series for signal (A & C) and noise (B & D) components from single-session ICA. ............................................................ 96

Supplementary Figure 2. Design and contrast matrices in FSL/GLM tool. ........................ 100

Supplementary Figure 3. Independent component maps of rs-fMRI group-ICA datasets overlaid on 3D-rendered standard Sprague Dawley brain atlas showing differences in resting activity in pre-stimulation, 10 Hz, BHFS, 1 Hz and cTBS groups. ........................................ 102

Supplementary Figure 4. Snapshots of the rat brain atlas corresponding to coordinates y=71 and z=46 (A) and y=60 and z=40 (B) in the pre-stimulation group-ICA components 0 and 11 respectively. ....................................................................................................................... 105

Table 1. Total number of pulses delivered during 10 min LI-rTMS stimulation at the different frequencies used. ................................................................................................................. 42

Table 2. MRI parameters used in the RARE anatomical scan and the functional T2* scan using single-shot gradient-echo echo planar imaging (EPI). ......................................................... 43

Table 3. Summary of changes in the default mode network in brain regions post-rTMS using 10 Hz and BHFS excitatory frequencies and 1 Hz and cTBS inhibitory frequencies on the right (R) side (ipsilateral changes) and on the left (L) side (contralateral changes) of the brain. 57

Table 4. Smoothing increased the total number of components and decreased the percentage of useful ICA components obtained from single-session MELODIC on the pre- and post- 10 Hz rs-fMRI datasets for the six rats. ................................................................ 59

Table 5. Reported statistics in recent rs-fMRI rodent studies. ............................................ 68

Supplementary Table 1: Examples of recent fMRI studies in rodents and humans and the respective fMRI data voxel size and smoothing used in pre-processing. ............................ 95

Supplementary Table 2. Comparison of group-ICA components when limiting the total number of components to 15 vs. 30 on the post- 10 Hz stimulation group. ....................... 98

Supplementary Table 3. Homologous ICA components (0 to 14) showing the DMN in the pre-stimulation group and the five post-stimulation groups (0 Hz, 10 Hz, BHFS, 1Hz and cTBS). .................................................................................................................................... 99


LIST OF ABBREVIATIONS

AMPA alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

BHFS biomimetic high frequency stimulation

BOLD blood-oxygen-level dependent

CBF cerebral blood flow

cTBS continuous theta burst stimulation

DMN default mode network

ECT electroconvulsive therapy

EPI echo planar imaging

fMRI functional magnetic resonance imaging

GFAP glial fibrillary acidic protein

Hz Hertz

ICA independent component analysis

iTBS intermittent theta burst stimulation

LI-rTMS low-intensity rTMS

LTD long-term depression

LTP long-term potentiation

M1 primary motor cortex

MEP motor evoked potential

MR magnetic resonance

MRI magnetic resonance imaging

NMDA N-methyl-D-aspartate

PET positron emission tomography

RARE rapid acquisition with relaxation enhancement

rs-fMRI resting state functional magnetic resonance imagining

rTMS repetitive transcranial magnetic stimulation

T Tesla

TBS theta burst stimulation

TMS transcranial magnetic stimulation


ACKNOWLEDGEMENTS

My honours year would not have been as pleasing and rewarding if it was not for the

exceptional group of people who have supported me throughout the year.

First and foremost, I would like to thank my principal supervisor Dr Sarah Etherington and

my co-supervisors A/Prof Jennifer Rodger and Dr Kirk Feindel, who provided me with

excellent direction and unwavering support since the start of the semester. My greatest

thanks go to Sarah who gave me valuable feedback on my written work, provided me

guidance in “academic social behaviour” and would arrange to meet me on only a day’s

notice sometimes. I owe a tremendous amount to Jenny for her clear perspectives and swift

feedback on my written work and also for training me in animal handling and the use of

rTMS technology. I am also very thankful to Kirk who taught me most of the research skills

I learnt this year - from laboratory techniques such as performing fMRI scans to data

analysis. It is mostly because of him that I was able to develop independence in the

laboratory and confidence in my abilities. I am fortunate to have had such considerate and

hard-working supervisors.

Thank you to my examiners Dr Ann-Maree Vallence and Dr Kristen Nowak for their helpful

feedback. I would also like to acknowledge Dr Andrew Mehnert for his help with the co-

registration and data analysis of the fMRI data and the scientific and technical assistance of

the National Imaging Facility at CMCA, UWA.

A special thanks to Ms Clare Auckland and Ms Marissa Penrose for driving all the way to

CMCA at 8.00am to assist me during the experiment sessions.

I thank the staff and students of the Experimental and Regenerative Neurosciences (EaRN)

laboratory (School of Animal Biology, UWA), who were always generous in their offers of

assistance with any matters.

Last but not least, I would like to thank my whole family for their support; specially, my

younger sisters, Deepty and Dishty Seewoo, for their encouragements to “hang in there”,

and my parents, Tookraj and Bharatee Seewoo, for always being there for me and for

keeping me motivated.


CHAPTER ONE: INTRODUCTION

1.1 OVERVIEW

Electrical stimulation of nerves, for example in electroconvulsive therapy (ECT), has been

shown to modulate depression and many types of seizures (Fujiki and Steward 1997). In

ECT, the excitable membranes of neurons are depolarised by applying a current to the body

via surface or implanted electrodes (George and Wassermann 1994). Even though this

technique has been widely used in the treatment of several psychiatric conditions, it can be

painful as the current can cause sensory neurons to fire (Barker and Freeston 2007).

Moreover, ECT can have undesirable side-effects such as amnesia and it is invasive as it

involves delivery of electrical signals via implanted electrodes (Fujiki and Steward 1997). It

is also difficult to stimulate deep brain structures via surface electrodes because of the high

electrical resistance of the skull (Barker and Freeston 2007).

An alternative non-invasive and painless approach is to use transcranial magnetic

stimulation (TMS) to induce a current in the brain, where it can cause neurons to fire by

changing their electrical environment (Wassermann and Zimmermann 2012). TMS is able to

stimulate the human brain and deep peripheral nerves without causing pain as there is no

induced current that passes through the skin, where most of the pain fibre nerve endings

are located (Barker and Freeston 2007). This lack of discomfort enables the technique to be

readily used on patients and volunteers for research and therapeutic purposes (Barker and

Freeston 2007).

Delivering long trains of closely spaced pulses (repetitive TMS or rTMS) can have effects on

neuro-excitability and behaviour that exceed the duration of the stimulation. For example,

rTMS delivered at low intensities (LI-rTMS) can promote re-organisation of abnormal or

damaged neural circuits in animals (Makowiecki et al. 2014) and cause long-term

behavioural changes in patients with psychiatric and neurological disorders (Pridmore and

Belmaker 1999). George and colleagues (1995) were the first to use rTMS as a tentative

treatment for depression. Since then, rTMS has been shown to have therapeutic potential

for a range of psychiatric disorders, including unipolar (Xia et al. 2008, Gaynes et al. 2014)

and bipolar depression (Xia et al. 2008), schizophrenia (Dlabač-de Lange et al. 2010),


obsessive-compulsive disorder (Jaafari et al. 2012) and post-traumatic stress disorder (Clark

et al. 2015) as well as neurological disorders such as Parkinson’s disease (Arias-Carrion

2008), dystonia (Machado et al. 2011), tinnitus (Soleimani et al. 2015), epilepsy (Pereira et

al. 2016) and stroke (Corti et al. 2012). rTMS has also shown promising results in the

treatment of pain syndromes such as migraine (Lipton and Pearlman 2010) and chronic pain

(Galhardoni et al. 2015). In 2008, an rTMS device developed by Neuronetics was approved

by the Food and Drug Administration in the United States for the treatment of patients with

major depressive disorder who are resistant to at least one antidepressant drug (O’Reardon

et al. 2007).

Even though rTMS is being used in a clinical setting and clinical trials are abundant, not much

is known about the mechanisms underlying its efficacy (Wassermann and Lisanby 2001).

This knowledge gap is because human studies use mostly non-invasive methods such as

fMRI, TMS and behaviour to investigate the effects of rTMS while animal studies mostly use

invasive methods. Animal models using the same outcome measures that are applied in

humans will allow for direct comparison of animal and human data and hence, provide

valuable insight into how the effects of rTMS in rodents might translate to humans. de

Sauvage et al. (2008) assessed the applicability of using rats to investigate the effects of

rTMS and extrapolating the findings to humans by comparing human and rat data in the

literature. The authors compared the current densities observed in the motor cortex in

human and rat brains at motor threshold. The results supported the use of rat models of

rTMS as the current densities at motor threshold in rats using flat custom-made coils did

not differ significantly from those in humans despite the difference in brain size and

structure (Section 1.2.5). de Sauvage et al. (2008) also evaluated the genotoxicity of rTMS

and showed that rTMS, when applied in a similar way as human rTMS during treatment of

depression, does not cause any damage to DNA in the rat brain cells.

Animal models can be used to study the effects of rTMS using non-invasive techniques as in

humans to allow direct comparison of human data with animal data and also with

subsequent invasive techniques to explore those effects in greater detail. This can

potentially impact on our knowledge of the underlying mechanisms by which rTMS has its


effects and hence allow for the development of optimal combinations of rTMS parameters

(e.g. frequency, intensity), which will improve the efficacy of rTMS as a therapeutic tool.

1.2 TRANSCRANIAL MAGNETIC STIMULATION (TMS)

1.2.1 The principle behind TMS

Faraday’s law of electromagnetic induction states that moving magnetic fields can induce a

transient electrical current in a nearby conductor (Faraday 1965). TMS is a technique that

has applied this principle to induce an electrical current in adult human brains (Barker et al.

1985, Walsh 1998). A wound coil is placed on the scalp of the subject over the region of

interest (Figure 1). The magnetic field, induced by the current flowing through the TMS coil,

passes into the brain unimpeded by skin, muscle or skull (Walsh 1998). This magnetic field

induces a current within the brain (Walsh 1998). Magnetic field strength is typically

expressed in Tesla (T). A classic TMS device creates a moderately powerful magnetic field of

1 to 2.5 T that lasts for a few milliseconds (Rossini et al. 1994).

Figure 1. Neuronal activation by TMS using "figure-of-eight" coil. The electrical current in the

coil (1) generates a magnetic field (2), which induces a current in the brain (3). This causes stimulation of


neurons (4), with the optimum site of activation shown as dark blue spots in the inset. The electrical current in the coil and the current induced in the brain flow in the same plane, tangential to the skull-brain surface. (Edwards et al. 2008).

TMS stimulates activity in neurons, affecting the functioning of the cortex (Walsh 1998). The

effects of TMS on neural function are then measured indirectly, for example by recording

muscle activity in the thumb (Walsh 1998). When TMS is directed over the primary motor

cortex (M1), the corticospinal tract is activated, causing the spinal cord and peripheral

muscles to produce neuroelectrical signals known as the motor-evoked potentials (MEPs)

(Figure 2) (Miniussi and Rossini 2011) and these outcome measurements will be discussed

in greater detail in section 1.3.1).

Figure 2. Transcranial magnetic stimulation (TMS) applied over the motor cortex activates neurons leading to a corticospinal volley down the corticospinal tract. Action potentials lead

to a contraction in the contralateral muscle evoking a motor-evoked potential (MEP) on electromyography (EMG). The peak-to-peak amplitude is used to estimate corticospinal tract excitability. Adapted from Klomjai et al. (2015).


Early TMS devices emitted only a single, brief pulse. Modern devices can generate a rapid

succession of pulses, called rTMS. These devices can discharge on and off for several

minutes. They can deliver pulses at different rates (frequency) and in different patterns

(Section 1.2.2.1).

The after-effects of rTMS depends on the experimental design (for review, see Klomjai et al.

2015). The factors that determine the magnitude and duration of change in excitability of

the brain and whether this change is an increase or a decrease can be divided into three

principal categories — geometry (coil shape and orientation), strength of stimulation

(intensity) and timing (Pell et al. 2011). Geometry refers to the interaction of the spatial

distribution and orientation of the induced electric field with the cortical neuroanatomy.

The intensity of rTMS used determines whether there is an action potential firing following

stimulation or not, with only high-intensity stimulation being able to trigger firing of an

action potential and hence a muscle response if applied to M1 (Section 1.2.2.2). Timing

refers to the direct influence of the timing of the stimulus pattern, for example, the

frequency of the stimulus and the duration of stimulation on the effect of rTMS on neuro-

excitability (Section 1.2.2) (Pell et al. 2011). In this study, our main focus is on the effect of

rTMS frequency.

1.2.2 Different rTMS paradigms

1.2.2.1 The different rTMS frequencies and patterns of stimulation

The effect of the different rTMS paradigms used depends on the inter-pulse interval

(number of pulses per second) and the pattern of stimulation (presence and length of

intervals between trains of pulses) (Oberman 2014). The four commonly used paradigms

are 1 Hz, 10 Hz, continuous theta burst stimulation (cTBS), and intermittent theta burst

stimulation (iTBS) (Figure 3). A relatively new but more complex stimulation pattern called

biomimetic high frequency stimulation (BHFS) has also recently been used in pre-clinical

studies (Rodger et al. 2012, Makowiecki et al. 2014) (Figure 3).


Figure 3. Repetitive transcranial magnetic stimulation (rTMS) protocols. Simple frequencies

(1 Hz and 10 Hz) consist of identical stimuli spaced by an identical inter-stimulus interval. Theta burst stimulation (TBS) involves bursts of high-frequency stimulation (3 pulses at 50 Hz) repeated with an interval of 0.2 s (5 Hz). In continuous TBS (cTBS), bursts are applied continuously for 40 s (i.e. 600 stimuli) without breaks. In an intermittent TBS (iTBS) protocol, bursts are delivered for 2 s, then repeated every 10 s (2 s of TBS followed by a break of 8 s) for a total duration of 190 s (i.e. 600 stimuli). Biomimetic high frequency stimulation (BHFS) involves 62.6 ms trains of 20 pulses repeated at a frequency of 9.75 Hz. Adapted from Klomjai et al. (2015).


Simple frequencies like 1 Hz (1 pulse per second) and 10 Hz (10 pulses per second) are used

to deliver regular continuous stimulus trains with no inter-train interval (Oberman 2014). A

large body of literature shows that low-frequency (1 Hz) rTMS decreases brain excitability

while high-frequency (10 Hz) rTMS increases brain excitability in humans (for review, see

Klomjai et al. 2015). Studies investigating MEP size after application of different rTMS

frequencies to M1 found that repeated high-intensity stimulation at a low frequency (1 Hz)

caused a transient reduction in cortical excitability, leading to a reduction in MEP amplitude

(Pascual-Leone et al. 1994, Chen et al. 1997). At higher frequencies (5-20 Hz), rTMS

enhanced the cortical responses, which was detected by an increase in MEP-size (Pascual-

Leone et al. 1994). Similar results were obtained when changes in absolute regional cerebral

blood flow (CBF) in adult patients with major depression were measured using positron

emission tomography (PET) imaging, following high and low frequency rTMS on the left

prefrontal cortex (Speer et al. 2000). 1 Hz stimuli decreased regional CBF in a small areas

within the right prefrontal cortex, left medial temporal cortex, left basal ganglia, and left

amygdala while 10-20 Hz rTMS stimuli caused a significant increase in the regional CBF in a

number of brain regions (prefrontal cortex, cingulate gyrus, left amygdala, bilateral insula,

basal ganglia, uncus, hippocampus, parahippocampus, thalamus, and cerebellum) (Speer et

al. 2000).

Theta burst stimulation (TBS) uses a composite stimulation pattern, consisting of repeating

bursts of stimuli (Larson et al. 1986). Each burst consists of 3 pulses of stimulation at 50 Hz,

and the bursts are repeated at 5 Hz (0.2 s) (for review, see Cardenas-Morales et al. 2010).

This pattern of stimulation is based on the endogenous brain oscillations observed in the

hippocampus (Huang et al. 2005). The effect of TBS on M1 depends on the pattern of

stimulation (Figure 3): cTBS or iTBS (Ljubisavljevic et al. 2015). The two main modalities

tested have been shown to have opposite effects on brain excitability. When a continuous

train of TBS was applied for 40 s (i.e. 600 stimuli) to M1 in humans (cTBS), it caused inhibition

of the corticospinal output, which was detected by a decrease in MEP amplitude (Figure 4)

(Huang et al. 2005). There were changes in intracortical facilitation and a short duration of

intracortical inhibition, which confirmed that TBS affects the excitability of neuronal circuit


within M1 (Huang et al. 2005). Intracortical facilitation is an excitatory phenomenon and

short interval intracortical inhibition is an inhibitory phenomenon evoked in the M1 by

conditioning stimuli (in this case, TBS) as detected in the EMG response. In contrast, when

TBS was applied with short breaks (iTBS), with a 2 s train of TBS repeated every 10 s (2 s TBS

followed by 8 s pause) for 190 s (i.e. 600 stimuli), a significant increase in MEP size (Figure

4) and short duration of intracortical inhibition was observed while intracortical facilitation

was unaffected (Huang et al. 2005). Huang et al. (2005) also showed that a shorter

stimulation time and a lower intensity application is required when using TBS over M1 to

have the same significant after-effects as other rTMS protocols (e.g. simple frequencies like

1 Hz and 10 Hz) and that these after-effects can last for up to an hour after application

(Huang et al. 2005).

Figure 4. The size of motor-evoked potentials (MEPs) is smaller after continuous theta burst stimulation (cTBS) and more pronounced after intermittent theta burst stimulation (iTBS) compared to baseline. After cTBS is applied to the motor cortex, MEPs show decreased amplitude

for a period of 20–30 min, whereas iTBS causes an increase in the amplitude of MEPs for approximately the same amount of time. Adapted from Oberman (2014).

Even though there can be a wide range of frequencies and stimulation patterns, the effect

of frequency has not really been investigated beyond the four conventional frequencies

described above (1Hz, 10Hz, cTBS and iTBS) in humans. There is a lot of potential to discover

other frequencies, which can possibly have different neuronal effects. This could allow

stimulus frequencies to be selected in a logical way in order to produce particular effects in

the brain. This potential has been shown in animal models using BHFS at low intensity


(Rodger et al. 2012, Morellini et al. 2015). BHFS is a complex waveform (Martiny et al. 2010)

that replicates the patterns of nerve activation in muscles during exercise (Rodger et al.

2012) and is based on the patent PCT/AU2007/000454 of Global Energy Medicine. Like TBS,

it has the potential to integrate with the brain’s own endogenous firing patterns, and

therefore to be more powerful than more simple firing frequencies. It involves a total of 10

min stimulation period and comprises of 62.6 ms trains of 20 pulses, repeated at a frequency

of 9.75 Hz (Grehl et al. 2015). The trains are repeated at 6.71 Hz for 1 min (warm-up), 10.1

Hz for 8 min (treatment), and 6.26 Hz for the last minute (cooldown) (Rodger et al. 2012).

BHFS has been used in the treatment of joint injuries but not in the brain because this

stimulation frequency is new and has not been fully characterised as yet. Although it has

not been applied therapeutically in humans, animal research has shown that it causes

reorganisation of abnormal neural networks (Rodger et al. 2012, Makowiecki et al. 2014)

and induces climbing fibre reinnervation in denervated hemi-cerebellum (Morellini et al.

2015).

1.2.2.2 The different rTMS intensities

The majority of rTMS studies define the intensity of the stimulation as a percentage of the

resting motor threshold (Rossini et al. 1994). The resting motor threshold is the minimum

stimulatory intensity required to produce an MEP of a minimum of 50 μV (Lee et al. 2006).

For rTMS stimulation intensities around the resting motor threshold (Lee et al. 2006),

membrane depolarization caused by the electric field from rTMS generates action

potentials, leading to synaptic transmission and excitatory or inhibitory postsynaptic

potentials (described in section 1.3.1). On the other hand, if rTMS is used at very low

intensities, for example during LI-rTMS applied in animal models (Rodger et al. 2012,

Makowiecki et al. 2014) as performed in the present study, or experimentally in humans

(Capone et al. 2009, Rohan et al. 2014), the weak induced currents do not induce action

potentials, but nonetheless induce plasticity by other mechanisms (Section 1.2.3).

1.2.3 The after-effects of rTMS – Synaptic Plasticity (LTP and LTD)

Through the use of different frequencies of stimulation, rTMS allows for transient

modulation of neural excitability as described above (Section 1.2.2.1). More importantly,

these effects can outlast the stimulation period, leading to synaptic plasticity (for review,


see Miniussi and Rossini 2011). Plasticity is any lasting structural or functional change in

neuronal or network properties following a change in the body or the external environment

(Donoghue et al. 1996). Synaptic plasticity refers to the long-term changes in the pattern

and functions of synaptic connectivity and is closely related to the development of the

nervous system and important brain functions, including injury repair, learning and memory

formation (Wang et al. 2010). An important mechanism of synaptic plasticity follows the

Hebbian theory (Hebb 1949) in which there is a cause-effect relationship between the firing

of presynaptic and postsynaptic neurons. Hebb (1949) hypothesised that synaptic efficacy

is enhanced when the presynaptic and the postsynaptic neurons are activated

simultaneously. The repetitive use of a function, for example in reinforcement-associated

behavioural motor-training tasks, is known to induce learning via this mechanism as it

activates specific cortical circuitry (Miniussi and Rossini 2011). Repetitive brain stimulation

may induce plasticity via the same principles as it can generate a prolonged depolarising

response in both pre and postsynaptic neurons, thus strengthening the neural circuitry.

Two important and widespread forms of synaptic plasticity are long-term potentiation (LTP)

and long-term depression (LTD). LTP and LTD are synaptic processes that increase or

decrease neuronal excitability respectively (Bliss and Lomo 1973, Malenka and Bear 2004).

Experimental work on animals and human studies suggest that the long-term effects of

rTMS on brain excitability are related to plastic changes similar to LTP and LTD (animal

studies: Vlachos et al. 2012, Funke and Benali 2011, human studies: Beck et al. 2000,

Thickbroom 2007).

The plasticity induced by rTMS appears to be dependent on the activation of NMDA

receptors (N-methyl-D-aspartate receptors) (Rowland et al. 2005), which are central to

classical LTP and LTD mechanisms. Repeated stimulation of neurons results in glutamate

binding to the NMDA receptor, causing the opening of magnesium-blocked Ca2+ channels

and hence, allowing Ca2+ to flow into the postsynaptic cell (Figure 5). Intracellular Ca2+

concentration then determines the direction of plasticity downstream of NMDA receptor

activation. The “calcium hypothesis” of LTP and LTD states that whether plasticity leads to

LTP or LTD is determined by the degree of the increase in intracellular Ca2+, with a big rise


in Ca2+ due to strong depolarisation inducing LTP, and a smaller increase in Ca2+ due to weak

depolarisation inducing LTD (Artola and Singer 1993).

Figure 5. Intracellular signalling mechanisms for long-term potentiation (LTP) and long-term depression (LTD). Glutamate (green) binds to NMDA receptors, causing the repulsion of Mg2+ ions

and influx of Ca2+. In LTP, Ca2+ binds to the Ca2+/calmodulin complex, which ultimately leads to the phosphorylation of the alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor. This causes more AMPA receptors to be inserted into the membrane and encourages the influx of Na+ ions upon binding of glutamate. In LTD, calcium activates phosphatases, which dephosphorylate the AMPA receptor, resulting in its internalisation and a decrease in its permeability to Na+. Adapted from Mc Dermott (2015).

A high increase in the concentration of Ca2+ activates a calcium-calmodulin-dependent

kinase, which phosphorylates alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

(AMPA) receptors in cortical neurons (Malenka and Bear 2004). Direct phosphorylation of

existing AMPA receptors and insertion of additional postsynaptic AMPA receptors facilitate

permeation of sodium ions, leading to LTP (Malenka and Bear 2004). These events are

consistent with stimulation using high-frequency rTMS, leading to potentiation (Bliss and

Cooke 2011).


In contrast, a weak depolarisation of the postsynaptic cell induces a small increase in

intracellular Ca2+ via the activation of NMDA receptors (Malenka and Bear 2004). This

triggers a different cascade of events leading to the dephosphorylation of the AMPA

receptor by a phosphatase (Malenka and Bear 2004). Dephosphorylation leads to a

reduction in the number of AMPA receptors on the postsynaptic membrane due to

internalisation of these receptors and there is a decrease in their permeability to sodium

ions, resulting in LTD (Malenka and Bear 2004). If NMDA receptors are blocked by NMDA

receptor antagonists, low-frequency rTMS is unable to cause neuronal modulation (Muller

et al. 2014). Therefore, stimulation at low frequency (1 Hz) may have an LTD-like effect as

the absence of neuronal effect in the presence of NMDA receptor blockers suggest that this

form of cortical plasticity induced by rTMS is dependent on the NMDA receptors (Muller et

al. 2014).

1.2.4 Variability in rTMS effects in humans

The induction of cortical plasticity as described above has been the driving force behind

clinical studies of rTMS as a potential treatment of neurological and psychiatric conditions.

However, even in neurologically normal subjects the variability in their response to rTMS is

high (Maeda et al. 2000). Increasing evidence suggests that rTMS may not induce reliable

and reproducible effects, therefore limiting the current therapeutic usefulness of this

technology (for review, see Ridding and Ziemann 2010).

Figure 6 below summarises some of the factors inducing variability in rTMS effects, although

many are still unknown. The gender of the subject induces the smallest variability in rTMS

results. Females are slightly more responsive to rTMS than males (Ridding and Ziemann

2010). Regular aerobic exercise and diet can also modify plasticity. These factors are known

to increase learning and memory formation via several mechanisms including increased CBF

and neurotrophic factors (for review, see van Praag 2009). Additionally, the variation of

cortisol levels with respect to time of the day makes time an important determinant of the

degree of plasticity induced following brain stimulation. It has been reported that there is

an increased plasticity in M1 in the afternoon when cortisol levels are low (Sale et al. 2007,

2008). It is well established that there is a decline in neuroplasticity, learning and memory


with age in both healthy individuals and those with neurological and psychiatric conditions.

A decline in rTMS-induced plasticity with age has also been noted in both cohorts (Ridding

and Ziemann 2010). The use of neuropharmacological drugs influences plasticity induced by

non-invasive brain stimulation. For example, given that cortical synaptic plasticity depends

on NMDA receptors (Section 1.2.3), drugs such as dextromethorphan, which act as NMDA

receptor antagonists can prevent rTMS from inducing LTP-like and LTD-like plasticity

(Ridding and Ziemann 2010). One of the potential mechanisms by which rTMS induces

plasticity is believed to be by inducing gene expression (Sections 1.4.1 and 1.4.2). As such,

genetic polymorphisms in plasticity-regulating genes may have a great impact on the

response to treatment (Ridding and Ziemann 2010).

Considering the significant inter- and intra-individual variability in the after-effects induced

by rTMS, a better understanding of how rTMS affects the brain is required (Muller-Dahlhaus

and Vlachos 2013). The extent to which variability is introduced into the experiments can

be limited in animal studies by controlling for gender, age, diet, drugs, genetic background

and time at which experiments are carried out. Therefore, animal models of rTMS can

possibly overcome the confounding effects of variability in the data and help us dissect out

the relative contributions and possible impacts of these factors.


Figure 6. Relative magnitude (y-axis) and potential significance (x-axis) of determinants of plasticity induction by non-invasive brain stimulation in persistently determining the direction and magnitude of the induced plasticity. Potential significance is a rather subjective

approximation made by the authors of the likelihood of the various determinants to persistently impact on the direction and magnitude of the induced plasticity in future studies. A percentage change of zero on the y-axis denotes the level of plasticity that would be expected to be induced by non-invasive brain stimulation in a randomly sampled group of adults. Colours of bars (see legend) indicate the level of evidence from the number and consistency of available studies. Abbreviations: AG, agonist; ANT, antagonist; DA, dopamine; Ach, acetylcholine; NE, noradrenaline. (Ridding and Ziemann 2010).

1.2.5 rTMS delivery in humans vs. rodents

In humans, the "figure-of-eight" coil (Figure 7A) is used with the stimulation hotspot (shown

in dark grey) positioned over the target brain region in order to provide optimal focal

stimulation (Rodger and Sherrard 2015). Brain areas adjacent to the stimulated region also

receive some stimulation but at much lower intensity.


Figure 7. Summary of available and desired coils for rTMS in animal studies that are equivalent to those used in human studies. The coils are shown in black and the estimated stimulated

region is in grey. Panel A shows the area of stimulation when human “figure-of-eight” coils are used in humans, mice and rats. Panel B shows a smaller “figure-of-eight” coil used for animals. Panel C shows custom-made circular coils used in rodents. Panel D shows how an ideal small “figure-of-eight” coil would be able to provide focal stimulation in animals while maintaining a similar coil to target ratio as that used in humans. Adapted from Rodger and Sherrard (2015).

Using the standard, human-sized “figure-of-eight” coil in animals would stimulate the entire

brain of the rodent and also an appreciable portion of its body (Figure 7A). The hotspot is

no longer focal relative to the target, and the induced current is no longer contained within

the target brain region (Vlachos et al. 2012). This effectively decreases the efficiency of rTMS

and changes the properties of the induced current, rendering the animal models unreliable

and extrapolations to focal stimulation in humans invalid (Wassermann and Zimmermann

2012).

The smaller size of rodents leads to additional considerations in comparison to humans

when scaling down rTMS for rodents due to the interaction of size and magnetic field


(Wassermann and Zimmermann 2012). Smaller sized “figure-of-eight” coils (Figure 7B) can

be used. This improves focality to one hemisphere in rats but still has reduced efficiency of

induction (Rodger and Sherrard 2015). To circumvent this issue, one possible approach is to

construct appropriately sized coils for use in animal studies. Custom-made round coils

(Figure 7C) have been used to deliver focal stimulation in rodents as they are small enough

to stimulate one hemisphere in both mice and rats (Rodger et al. 2012, Makowiecki et al.

2014), with the induced current fully contained within the brain, increasing the efficiency of

induction. These coils induce the current as circular loops with the induced current

stimulation hotspot being under the windings of the coil (Rossini et al. 1994). There is no

induced current under the centre of the coil, and therefore, no stimulation occurs there

(Rossini et al. 1994).

However, there is a heat dissipation problem when using small round coils, especially with

high-intensity rTMS (Cohen and Cuffin 1991). Therefore, the custom-made coils can deliver

only low-intensity magnetic fields (LI-rTMS) (Rodger and Sherrard 2015). While most of the

animal studies in the literature (except for those conducted by our laboratory) have used

the large human coils in spite of the issues discussed above, in the present study, we have

used a small custom-made circular coil to deliver LI-rTMS to rats to one hemisphere of the

brain. Even though LI-rTMS cannot induce action potentials in the brain, it is able to induce

changes in intracellular calcium levels and regulate gene expression (Grehl et al. 2015),

which underpin short and long lasting changes in excitability and structural reorganisation

(Makowiecki et al. 2014) (Section 1.4).

1.3 MEASURING THE EFFECTS OF rTMS NON-INVASIVELY

1.3.1 Detecting the effects of rTMS using EEG and EMGs

Studies on the excitability of M1 in humans have revealed how the physiology of the cortex

is changed immediately following rTMS due to its activation. TMS-compatible

electroencephalography (EEG) equipment, an approach that is gaining in popularity, was

used by Komssi et al. (2004) to detect the effect of rTMS on neuronal activation. Stimulation

intensity of only 60% motor threshold could elicit a clear response in the brain as a distinct

global mean field amplitude waveform was detected in the EEG recordings in all subjects

(Komssi et al. 2004, Kähkönen et al. 2005).


A change in motor cortical excitability can also be detected as a change in the amplitude of

MEPs in peripheral muscles (Figure 2 & 4) following rTMS as compared to baseline MEP

amplitude (Wassermann and Zimmermann 2012). High-intensity TMS is used to deliver

stimulation above the resting motor threshold, and a twitch is observed in the peripheral

muscles (Lee et al. 2006). There is adequate membrane depolarization caused by the

electric field from TMS, which generates action potentials, leading to synaptic transmission

and excitatory or inhibitory postsynaptic potentials (Klomjai et al. 2015). The size of the

evoked muscle response detected as electromyographic (EMG) activity in specific muscles

reflects the excitability of the corticospinal system (see Figure 4 for the difference in MEP

size before and after cTBS and iTBS) (Lee et al. 2006).

1.3.2 Detecting rTMS-induced changes using brain imaging techniques

Several brain imaging procedures such as single photon emission computed tomography,

functional magnetic resonance imaging (fMRI) and PET have been used to study how brain

activity is modified by rTMS during or up to 30 min after stimulation in humans (Reithler et

al. 2011). The changes in brain activity using brain imaging techniques are determined from

a change in cerebral glucose metabolism and blood flow, which reflect the level of cellular

activity (Raichle 2009). Changes in haemodynamics in the brain are coupled with neural

activity and hence co-vary with stimulation duration and frequency of rTMS used (Allen et

al. 2007). Increases and decreases in activity have been found depending on the stimulated

brain region and the frequency of rTMS (Bohning et al. 1999, Kimbrell et al. 1999). For the

purpose of this study, we will focus on the use of fMRI to detect changes in brain activity.

1.3.3 fMRI study of brain activation

1.3.3.1 Background to MRI

MRI provides mainly anatomical information (Ogawa et al. 1990). It uses a strong magnetic

field, generated by a large magnet, to align the nuclear spins of the hydrogen atoms in

water. Since this magnetic field is static, the magnetic resonance (MR) signal received by

receiver coils has no spatial information and hence, cannot be used to create an image. A

gradient coil is used to give this spatial information by generating short-term spatial

variations in magnetic field strength across the subject. Transmitter coils send radio


frequency pulses, which create a secondary magnetic field that oscillates perpendicular to

the main field. This secondary magnetic field exerts a torque on the net magnetisation that

causes it to rotate out of alignment with the scanner’s main magnetic field, and into the

plane in which the secondary magnetic field oscillates. The net magnetisation rotating

induces a voltage and current in the receiver coil. After a few milliseconds, the nuclear spins

dephase due to interactions with their surroundings and the signal intensity decreases.

Figure 8. Anatomical MRI image of a healthy human brain showing the left (1) and right (2) dorsolateral prefrontal cortex and left anterior cingulate cortex (3) in trans-axial orientation. Adapted from Michael et al. (2003).

While MRI is static, there is an extension of this technique that is used to measure brain

activity: fMRI is a neuroimaging procedure that uses a powerful magnetic field with radio

frequency pulses to non-invasively measure brain activity by detecting changes in blood

flow and oxygenation in the draining veins, that is, downstream of the active brain region

(McRobbie et al. 2007, 340). The basis of fMRI is that it measures the haemodynamic

response to change in neuronal activity based on the change in the ratio of oxyhaemoglobin

and deoxyhaemoglobin in the blood (Ogawa et al. 1990). Physiological activation of the

brain leads to an increase in CBF in that specific brain region. This increase is, however,

greater than the increase in demand for and hence the supply of oxygen (Mintun et al. 2001)

or glucose (Powers et al. 1996) to neurons. The increase in CBF, therefore, does not match

the increase in oxygen consumption by the brain tissue (Fox and Raichle 1986, Fox et al.


1988). As such, brain activation increases blood oxygenation in blood vessels in the brain

(Figure 9) (Ogawa et al. 1990).

Figure 9. The difference in blood oxygenation at resting state (a) and when the brain is

active (b). Over-provision of fully oxygenated blood in active brain regions causes a reduction in the

concentration of deoxyhaemoglobin in draining veins. (McRobbie et al. 2007)

Blood oxygenation is an important factor that determines the magnitude of the fMRI signal.

Deoxyhaemoglobin in the blood has been shown to change the signal from water molecules

surrounding a blood vessel (Ogawa et al. 1990). The iron in haemoglobin, due to its

electrons, acts as a tiny magnet when exposed to a magnetic field during an fMRI scan

(Raichle 2009). The effect of the iron is masked when the haemoglobin is oxygenated, such

that oxyhaemoglobin does not disrupt the magnetic field (Raichle 2009). Oxyhaemoglobin

is said to be diamagnetic (only paired electrons) but when it loses its oxygen molecules, it

becomes deoxyhaemoglobin, which is paramagnetic (has unpaired electrons) due to the

exposed iron (Ogawa et al. 1990, Raichle 2009). Para-magnetism affects magnetic

susceptibility, which means that the presence of paramagnetic deoxyhaemoglobin

decreases the signal intensity and creates a contrast between the blood vessel and the


surrounding tissue (Ogawa et al. 1990). This contrast depends primarily on blood oxygen

level and is therefore controlled by the ratio of blood supply or flow and demand by the

tissue. The contrast created by the presence of deoxyhaemoglobin can be used to monitor

blood oxygenation in different parts of the brain based on metabolism or blood flow (Ogawa

et al. 1990). This is more commonly referred to as blood-oxygen-level-dependent contrast

imaging or BOLD-contrast imaging. Both an increase and a decrease in brain activity from a

physiological baseline can be detected in fMRI because it uses the paramagnetism of

deoxyhaemoglobin as an in vivo MRI contrast agent (Raichle 2009).

1.3.3.2 Use of fMRI to determine changes induced by rTMS

Bohning et al. (1998) were the first to combine the use of TMS and fMRI in human subjects.

They applied high-intensity TMS to the brain, which resulted in a significant increase in

activity in M1 as detected by the fMRI scan. Their experiment demonstrated the feasibility

of interleaving TMS and fMRI protocols, that is, simultaneous TMS stimulation and fMRI

imaging inside an MRI scanner. This approach allows for direct visualisation of TMS-induced

changes in brain activity.

Soon after, the same authors compared the effect of different intensities of TMS on the

pattern of activation (80% and 110% resting motor threshold) of neurons in the brain and

reported an increase activity in both the local brain region (M1) and also in areas distal to

the stimulation site (for example the contralateral motor cortex and ipsilateral cerebellum),

illustrating the promise of this technique for mapping patterns of connectivity between

brain areas (Bohning et al. 1999). They also showed that high-intensity (above motor

threshold) TMS was associated with a greater increase in activity both locally and remotely

compared to low-intensity (less than motor threshold) TMS. However, the increase in

activity could be the result of cortical processing of re-afferent feedback from contractions

evoked in the hand muscle by the supra-threshold stimulation rather than the result of a

direct effect of TMS itself. This hypothesis was then supported by Bestmann et al. (2004),

who used interleaved fMRI/rTMS to compare the activation pattern for high- and low-

intensity rTMS. They showed that fMRI is able to detect the effects of sub-threshold

stimulation, thus validating our approach of using fMRI to detect effects of LI-rTMS.


According to Bestmann et al. (2004), the changes detected in brain regions remote to the

site of stimulation when LI-rTMS is applied to the cortex are due to altered synaptic

transmissions remotely caused by a change in excitability of the stimulated region.

Therefore, fMRI can be used to assess connectivity before and after rTMS to study the

changes induced by stimulation. A range of human studies has been conducted, which have

combined the use of rTMS with fMRI in order to record the rTMS-induced changes in

hemodynamic activity both in healthy subjects and subjects with neurological diseases

(Schneider et al. 2010, Fox et al. 2012). In healthy subjects, for example, fMRI was used to

detect plastic changes induced in the brain when 5 Hz LI-rTMS was applied to the right

dorsolateral prefrontal cortex (Esslinger et al. 2014). No change in activation was detected

at the stimulation site, but there was increased connectivity to the ipsilateral superior

parietal lobule, which is functionally connected to the right dorsolateral prefrontal cortex

during working memory (Esslinger et al. 2014). This increased connectivity was associated

with a decrease in reaction time during a working memory task (n-back task). These results

indicated the presence of rTMS-induced plasticity in prefrontally connected networks

downstream of the stimulation site (Esslinger et al. 2014).

Given that the pathophysiology of many psychiatric and neurological disorders is believed

to be related to altered neural connectivity and network dynamics, interleaved fMRI/rTMS

protocols provide an opportunity to investigate altered patterns of neural activity in these

disorders (for review, see Hampson and Hoffman 2010). Firstly, the activation patterns in

healthy individuals and patients with neurological or psychiatric conditions can be

compared after an rTMS session in order to determine how those patterns are disrupted in

the diseased state (Hampson and Hoffman 2010). For example, Schneider et al. (2010)

examined the effect of 5 Hz rTMS on the primary somatosensory cortex in dystonia patients

(a condition associated with impaired somatosensory ability) and healthy controls based on

their ability to discriminate between two stimulation frequencies applied to the right index

finger before and after the rTMS session. An fMRI scan was carried out together with the

tactile discrimination task. rTMS led to an improved performance in the tactile

discrimination task in healthy controls but not in dystonia patients (Schneider et al. 2010).


There was increased activity detected in the stimulated primary somatosensory cortex and

bilateral premotor cortex in both groups but fMRI detected an increase in activity in the

basal ganglia in healthy subjects only (Schneider et al. 2010). The authors hypothesised that

this could be related to altered sensory circuits and sensorimotor integration in dystonia

patients.

Secondly, the change in connectivity within the brain brought about by the use of rTMS as

a treatment method can be used to determine the neural mechanisms of improvement in

symptoms in the patients. For example, a longitudinal study by Gonzalez-Garcia et al. (2011)

used fMRI to investigate the mechanisms by which 25 Hz LI-rTMS (10 trains of 100 pulses)

over M1 for a period of 3 months improves the motor symptoms of Parkinson’s disease. It

was found that rTMS caused an increase in activity in the caudate nucleus during a simple

motor task (finger tapping experiment). fMRI also showed a decline in activity in the

supplementary motor area, which was accompanied by an increase in its functional

connectivity to the prefrontal areas (Gonzalez-Garcia et al. 2011). These changes

substantiated the beneficial effect of rTMS on the symptoms of Parkinson’s disease

observed in the patients.

Combining fMRI and rTMS has proven useful in a range of other ways. For instance, it has

provided evidence supporting the link between neural activity in the left inferior prefrontal

cortex and episodic memory formation (Köhler et al. 2004), demonstrated the involvement

of premotor cortical areas in speech perception (for review, see Iacoboni 2008) and enabled

targeting of stimulation based on the individual’s MRI anatomy and hence improvement in

rTMS effects (Fitzgerald et al. 2009). Because rTMS can have a potential impact in a range

of conditions and the induced changes can be detected by fMRI, animal models combining

these two techniques are warranted in order to uncover the mechanisms by which rTMS

can have this wide variety of effects.

1.3.3.3 Resting-state fMRI, the default mode network and rTMS

Resting-state fMRI (rs-fMRI) is used to detect functionally linked brain regions whose

patterns of spontaneous BOLD contrast fluctuations are temporally correlated when the

subject is at rest, that is, when no specific stimulus or task is presented (Biswal et al. 1995).


These fluctuations are thought to reflect the intrinsic functional architecture of the brain.

Brain regions with coherent spontaneous fluctuations in activity form an organised network

called the resting-state network (Biswal et al. 1995). The default mode network (DMN) is

one of the resting-state networks with a synchronised activity pattern. It shows the highest

activation when the subject is at rest and is deactivated in goal-oriented tasks (Raichle et al.

2001). When the subject is at rest, the DMN has a high glucose metabolism (Minoshima et

al. 1997) and regional blood flow (Raichle et al. 2001) compared to other brain regions. The

DMN has been associated with cognitive performance and is thought to play an important

role in neuroplasticity through the consolidation and maintenance of brain function

(Marcotte et al. 2013). For example, a higher resting-state activity within the DMN would

favour network efficiency (Kelly et al. 2008) while decreased connectivity between the

frontal and posterior DMN brain regions would result in functional deficits (Davis et al.

2009).

Compared to healthy individuals, people with neurological and psychiatric disorders have

been identified with DMN dysregulation (for review, see van den Heuvel and Hulshoff Pol

2010). Disruptions in functional connectivity between brain regions forming part of the

DMN have been implicated in, inter alia, conditions like Alzheimer’s disease (Greicius et al.

2004), multiple sclerosis (Lowe et al. 2002, Sbardella et al. 2015), autism (Cherkassky et al.

2006, Kennedy et al. 2006), epilepsy (Waites et al. 2006), depression (Greicius et al. 2007),

schizophrenia (Bluhm et al. 2007, Whitfield-Gabrieli et al. 2009), aphasia (Marcotte et al.

2013) and addiction (Sutherland et al. 2012, Lerman et al. 2014). It has been shown that

rTMS is able to modulate the resting-state activity of the brain and that DMN plasticity is

sensitive to rTMS in humans but the direction (increase or decrease in activity) and extent

of this modulation depend on the rTMS protocol used (for review, see Fox et al. 2012, more

recent articles: Popa et al. 2013, Glielmi et al. 2014, Jansen et al. 2015, Li et al. 2016).

Interleaving rs-fMRI and rTMS has opened doors to many possibilities in the clinical setting.

However, there have been no reports of animal studies using those same techniques and

also, no systematic investigation of LI-rTMS-induced plasticity using rTMS protocols

comparable to those used in humans. Because rodents are widely used as preclinical models


of various neuropsychiatric diseases, a thorough understanding of how LI-rTMS affects the

rodent DMN is of particular importance for both interpreting rodent rs-fMRI data and

translating findings between animal models and humans. Rs-fMRI studies in rodents have

successfully shown that rodents possess a DMN similar to humans despite the distinct

evolutionary paths between rodent and primate brains (Lu et al. 2012). Rs-fMRI, being a

task-free technique, allows for the investigation of connectivity without relying on

behaviour and therefore, can be applied to anaesthetised animals.

The present study aims to investigate whether LI-rTMS, applied to one hemisphere of the

brain, alters the strength or the spatial distribution, or both, of the DMN activity in rats.

Information about the extent to which DMN activity can be modulated by different rTMS

frequencies may prove helpful in the development of treatment options to modify

functional connectivity abnormalities based on the specific brain regions in different disease

states with dysfunctional DMN activity. To our knowledge, this is the first study to use rs-

fMRI to examine the effects of rTMS in rats and a brief summary of the relevant literature

will put this study in the context of what is already known from animal studies of rTMS.

1.4 MEASURING THE EFFECTS OF rTMS IN ANIMALS (INVASIVE)

1.4.1 Induction of the expression of plasticity-related genes

TMS can affect normal cellular function by inducing intense neuronal activity, which

influences gene expression within the brain cells (Pridmore and Belmaker 1999). As shown

in humans using imaging techniques, when rTMS is applied to the cortex, its effects are

distributed across the whole brain depending on the frequency used (Section 1.3.3). These

effects can be detected and studied in animals by measuring changes in gene expression in

different brain regions, which provide a compelling view of how rTMS can affect areas of

the brain remote to the site of stimulation.

Fujiki and Steward (1997) showed that high-frequency rTMS (25 Hz for 10 s) can up-regulate

the expression of the glial fibrillary acidic protein (GFAP) in mice. GFAP is the main

intermediate filament in mature astrocytes and is important for modulating astrocyte

motility and shape by providing structural stability to extensions of astrocytic processes.

The GFAP mRNA level was elevated for up to 8 days after treatment, with its peak at 24


hours post stimulation (Fujiki and Steward 1997). Even though the whole brain received the

stimulus, its effect was more pronounced in the hippocampal dentate gyrus (Fujiki and

Steward 1997). The authors hypothesised that the preferential increase in GFAP expression

in this specific brain region could reflect the pattern of neural activity generated by the

stimulation parameters used (10 trains of 250 pulses delivered at 25 Hz for 10 s). If this

hypothesis holds, the use of different stimulus parameters may induce a different pattern

and magnitude of neural activation and can, therefore, target different parts of the brain,

not just the dentate gyrus (Fujiki and Steward 1997). Ji and colleagues (1998) carried out a

study using high-frequency rTMS in rats to map different activated brain regions by tracking

the expression of immediate-early genes in rat brains. These genes are transcription factors,

which control the expression of other genes directly (Fujiki and Steward 1997). Using in-situ

hybridisation, a single application of 50 rTMS pulses delivered at 25 Hz for 2 s was found to

induce the expression of c-fos mRNA in distinct brain regions (Ji et al. 1998). This was

confirmed by immunocytochemical staining, which detected an increase in the level of c-

Fos and c-Jun proteins (Ji et al. 1998).

From these two experiments, it can be hypothesised that the thresholds for activation of

GFAP and c-fos genes are different, which means that the use of different protocols can

stimulate transcription of different genes (Pridmore and Belmaker 1999). Confirming the

specificity of gene regulation, rat models of stroke showed that changes in gene expression

induced by rTMS were mainly dependent on frequency and pattern of stimulation and were

associated with improved recovery from stroke (Ljubisavljevic et al. 2015). Only iTBS

induced significant changes in gene expression, while there were no significant changes

when stimulation was carried out at 1 Hz, 5 Hz and cTBS (Ljubisavljevic et al. 2015). Similarly,

treatment with iTBS for two weeks resulted in a significantly superior improvement in

neurological deficits as detected by behavioural testing scores compared to the other rTMS

protocols, suggesting that iTBS is able to promote repair after cerebral ischemic-reperfusion

injury (Ljubisavljevic et al. 2015). The difference in the effectiveness of the different rTMS

protocols may be related to their specificity in cell-type on which they have their effects.

None of the rTMS protocols caused downregulation of any genes and the 52 upregulated


genes due to iTBS were involved in angiogenesis, inflammation, injury response and cellular

repair, structural remodelling, neuroprotection, neurotransmission and neuronal plasticity

(Ljubisavljevic et al. 2015). Therefore, iTBS was shown to induce complex changes in gene

expression in the brain, which are able to promote recovery (based on behavioural testing

data) (Ljubisavljevic et al. 2015). Thus, there is evidence that using different rTMS paradigms

has the potential to selectively stimulate gene expression in neural and supportive elements

in regions of interest in different health conditions, such as in promoting recovery repair

after stroke (Ljubisavljevic et al. 2015).

1.4.2 Up-regulation of growth factor expression

Ljubisavljevic et al. (2015) identified brain-derived neurotrophic factor (BDNF) to be one of

the 52 up-regulated genes after iTBS in the rat model for stroke. BDNF is a member of the

neurotrophin family of growth factors, and neurotrophins like BDNF promote synaptic

plasticity (Waterhouse and Xu 2009), and are important for neuronal cell survival,

development, differentiation, and regeneration (Huang and Reichardt 2001). Because BDNF

is a secreted protein, it has the capacity to induce changes beyond the site of its expression

and may explain the widespread changes observed in the brain in humans using imaging

techniques such as PET and fMRI after treatment with rTMS (Section 1.3.3). Up-regulation

of BDNF is believed to play a crucial part in the molecular mechanism underlying LTP in the

hippocampus as it is released in an activity-dependant manner and has a significant role in

promoting changes in synaptic efficacy (Lu 2003, Bramham 2008). The effects of BDNF on

neural circuits and its trans-synaptic effects can explain the long-range impact of rTMS.

Korte et al. (1995) and Patterson et al. (1996) independently worked on BDNF knock-out

mice and reported an impairment of early LTP in BDNF homozygous or heterozygous

mice. Because rTMS studies had thoroughly described how rTMS may cause synaptic

plasticity via LTP-like mechanisms (Section 1.2.3), further studies explored whether rTMS

also affected BDNF expression in the brain. Both BDNF mRNA and protein levels have been

shown to significantly increase in rodent brains after rTMS or LI-rTMS stimulation,

suggesting that rTMS/LI-rTMS can promote BDNF expression in the brain (Müller et al. 2000,

Zhang 2007, Gersner et al. 2011, Rodger et al. 2012). Müller et al. (2000) reported that


several sessions of high-frequency rTMS (20 Hz) in awake rats resulted in a significant

increase in BDNF mRNA in the brain within 20 hours after stimulation. More recently,

Gersner et al. (2011) used western blot to detect the level of BDNF in rat brain 3 days after

the last stimulation session using high-frequency rTMS (20 Hz). They had similar results as

it was found that high-frequency rTMS in awake rats led to a long-term (3 days) increase in

hippocampal BDNF levels. They hypothesised that the induced increase in BDNF levels could

underlie the neuroprotective and neurorestorative effects of rTMS.

1.4.3 Changes in receptor expression

Regulation of NMDA and AMPA receptors have been proposed to underlie the LTP and LTD-

like mechanisms of rTMS (Section 1.2.3), leading to its therapeutic effects (Kole et al. 1999,

Gersner et al. 2011). Kole et al. (1999) delivered a single high-intensity rTMS session to

female Wistar rats at a stimulation frequency of 20 Hz for 3 seconds. The rats were sacrificed

24 hours after stimulation and the brain tissue was used for receptor binding studies. It was

found that there was a non-specific increase in NMDA binding sites in the hippocampal

formation of both experimental and control subjects and a specific increase in the number

of these receptors in the ventromedial hypothalamus, the basolateral amygdala and the

parietal cortex of the rats that received active stimulation. This showed that the effect of

rTMS is persistent (24 hours post stimulation) even after a single rTMS session and that the

increase in the number of NMDA receptors could underlie the persistent effect of rTMS.

Gersner et al. (2011) studied the effect of rTMS on the GluR1 subunit of AMPA receptors in

male Sprague Dawley rats. Phosphorylation of AMPA receptors occurs mostly at the site of

their GluR1 subunits, which increases the receptor’s permeability to calcium (Gersner et al.

2011). The rats received high-intensity brain stimulation using both high (9 trains of 100

pulses delivered at 20 Hz with an inter-train interval of 55 s) and low (1 Hz for 900 s)

frequency rTMS daily for 10 days. The rats were sacrificed 3 days after stimulation. High,

but not low, frequency rTMS caused a significant increase in GluR1 as well as the

phosphorylated form of GluR1 in the hippocampus of the rats. These results show that high-

frequency stimulation can induce the expression of receptors involved in LTP and LTD

mechanisms, which outlast the duration of the stimulation session.


1.5 PRESENT STUDY

The purpose of the current study was to use rs-fMRI in rats to investigate whether the effect

of LI-rTMS is localised and occurs directly at the site of stimulation or leads to changes in

brain regions remote to the site of stimulation as well. We have also explored the effects of

different frequencies of LI-rTMS stimulation using rs-fMRI. We have compared fMRI images

of rat brains before and up to 30 minutes after exposure to a train of LI-rTMS. All frequencies

(including controls) were tested in each rat, with at least a one-week interval between

frequencies to allow for any LI-rTMS effects to subside.

1.5.1 Hypotheses and Aims

Hypothesis 1: LI-rTMS will alter the resting-state activity of neurons directly at the site of

stimulation as well as in brain regions that have direct connections with the site of

stimulation.

AIM 1: To measure brain activation before and after LI-rTMS to one side of the brain.

Hypothesis 2: The magnitude and pattern of the change in resting-state neuronal activity

will depend on the frequency and pattern of LI-rTMS.

AIM 2: To establish the pattern of resting-state activity in the brain based on LI-rTMS

frequencies. Four frequencies were used (in addition to sham 0 Hz): 1 Hz, 10 Hz, cTBS, and

BHFS.


CHAPTER TWO: MATERIALS AND METHODS

2.1 ANIMALS

Six adult male Sprague Dawley rats (150-250 g) were sourced from the Animal Resources

Centre (Canning Vale, WA, Australia). They were maintained in a temperature-controlled

animal care facility on a 12-hour light-dark cycle with food and water ad libitum at M-block,

QEII Medical Centre. Experimental procedures were approved by the UWA Animal Ethics

Committee (RA/3/100/1430) and Murdoch Animal Ethics Committee (IRMA2848/16). All

investigators were trained in animal handling by the Programme in Animal Welfare, Ethics

and Science (PAWES) and had a valid Permission to Use Animals (PUA) license.

2.2 OVERVIEW OF THE EXPERIMENTAL DESIGN

The rats were allowed one week of habituation period upon arrival at the Animal Care Unit

before the start of the experiments. Rs-fMRI scans were performed immediately before and

after the stimulation session. Each rat received LI-rTMS stimulation at one of five

frequencies (0 Hz/sham, 1 Hz, 10 Hz, cTBS and BHFS, randomised order) in the morning each

week for five weeks (Figure 10). The timing of experiments was dependent on the

availability of imaging equipment but at least one week was allowed between sessions to

allow for any effect of LI-rTMS to subside (Li et al. 2013). Animals were kept for up to 12

weeks and were euthanised after the last rTMS stimulation session using carbon dioxide

asphyxiation.


Figure 10. Timeline of an experiment for one rat and protocol for each session. A. Timeline

for a single rat from the time of its arrival. The experiment consisted of an initial one-week period of

habituation upon arrival of the animal, after which the rats had 5 sessions of fMRI-LI-rTMS-fMRI, each

separated by at least a week, determined by equipment availability. Sessions 1 to 5 were the same, except for

the frequency of LI-rTMS used. B. Protocol for a single LI-rTMS session. During each session, the animal was

first anaesthetised using isoflurane gas and was kept under isoflurane anaesthesia throughout the experiment

(Section 2.3). Baseline rs-fMRI image was acquired after which sham/stimulation at a specific frequency (0 Hz,

1 Hz, 10 Hz, BHFS or cTBS) was delivered for 10 minutes to the right side of the brain by gently placing a circular

stimulation coil on the animal’s scalp (Section 2.4). A post-procedure rs-fMRI scan was then carried out. Each

of the six rats used in the experiment followed the same timeline, but the order in which the different

stimulation frequencies were delivered in the weekly experiments was randomised. The sections refer to the

part of the thesis in which that information is described.

2.3 ISOFLURANE ANAESTHESIA PROCEDURE AND MONITORING

The rats were pre-anaesthetised in an induction chamber using 4% isoflurane in 100%

medical oxygen delivered at 2L/min. Full anaesthesia was determined based on the absence

of a response to foot pinch. Once fully anaesthetised, the animal was transferred to a

heated imaging cradle and an isoflurane nose cone placed over the snout of the rat (Figure

11) that delivered 1-2.5% of isoflurane in 100% medical oxygen at approximately 1 L/min

inhalation.


Figure 11. Animal position on the imaging cradle during LI-rTMS stimulation and fMRI scans. Photography was limited due to the inability to bring magnetic objects into the imaging room.

After transfer to the cradle, some ophthalmic ointment was applied to both eyes to protect

the corneas from drying out while the animal was under anaesthesia. The animal’s

physiological conditions including body temperature, respiratory rate, heart rate, and

arterial blood oxygen saturation were monitored using a PC-SAM Small Animal Monitor (SA

instruments Inc., 1030 System). A rectal temperature probe with disposable lubricated

sleeve was used to monitor the rat’s temperature. A pneumatic pillow placed under the

rat’s abdomen was used to monitor respiration. Heart rate and blood oxygen saturation

were measured using an MR compatible pulse oximeter positioned at the base of the tail

(Figure 11).

2.4 LOW-INTENSITY REPETITIVE TRANSCRANIAL MAGNETIC STIMULATION

2.4.1 Coil Structure and Positioning

To induce stimulation, we used low-intensity magnetic field stimulation delivered using a

custom-built round coil (8 mm inside diameter, 16.2 mm outside diameter, 10 mm

thickness, 0.25 mm copper wire, 6.1 Ω resistance, 462 turns) placed on the right side of the

rat brain next to the right ear (Figure 12) (Grehl et al. 2015). The coil delivered a magnetic

field of 13 mT, which is below the motor threshold.


Figure 12. Stimulation coil positioning in relation to rat’s head and brain. The coil was placed

on the skull next to the ears, on the right side. Stimulation coil was connected to the pulse generator using a jack. Adapted from Rodger and Sherrard (2015).

The coil and pulse generator (Model EXLAB606, Serial Number 00003) were designed and

built by Global Energy Medicine (Perth, WA, Australia). The generator used was a version of

an e-cell, which is a commercially available electromagnetic pulse generator for therapeutic

treatment of musculoskeletal disorders. It is a simple resistor-inductor circuit under control

of a programmable (C-based code) micro-controller card (CardLogix, USA). The device was

modified such that custom designed coils could be attached (Grehl et al. 2015).

2.4.2 Stimulation Protocols

During each experiment, stimulation was delivered for 10 minutes at one of five

frequencies: 0 Hz, 1 Hz, 10 Hz, cTBS (3 pulses at 50 Hz repeated at 5 Hz) or BHFS (62.6 ms

trains of 20 pulses, repeated at 9.75 Hz). Each frequency had a specific pre-programmed

card such that when inserted into the generator, the phase transitions were triggered

automatically. The total number of pulses delivered for each stimulation paradigm is shown

in Table 1. For all experiments, controls were treated identically except during stimulation

when the coil was not activated (0 Hz). Sham stimulation controlled for any components of

the experimental procedure (handling, anaesthesia, positioning, and sound) that might

influence the fMRI signal.


Table 1. Total number of pulses delivered during 10 min LI-rTMS stimulation at the different frequencies used. Adapted from Grehl et al. (2015).

Frequency Total number of pulses delivered

0 Hz (sham) 0

1 Hz 600

10 Hz 6000

cTBS 7000

BHFS 120,000

2.5 MRI SCAN PROCEDURE

2.5.1 Equipment

The rats were scanned twice during the same session using the Bruker Biospec 94/30 small

animal MRI system operating at 9.4 T (400 MHz, H-1), with an Avance III HD console, and

BGA-12SHP imaging gradients. ParaVision 6.0.1 software was used to control the scanner

and set the experimental tasks. An 86 mm (inner diameter) volume coil was used to transmit

radio frequency pulses to generate an MRI signal, which was then detected using a rat brain

surface coil placed on the rat’s head. Surface coils have excellent sensitivity to signals from

nearby regions.

2.5.2 Image acquisition parameters

The acquisition parameters for fMRI were adapted from the studies by Tambalo et al.

(2015). A scout image (localising scan) was acquired to position the rat brain in the magnet

isocenter. Most structural and functional MRIs involve the reconstruction of three-

dimensional images from a series of two-dimensional slices. A spatial magnetic field

gradient and a radio frequency pulse are used to elicit a signal from a particular slice. In

practice, the slices are imperfect, and there may be some overlap of slice edges, which can

result in saturation of the MR signal. To minimise this issue, the slices were excited in a non-

contiguous, interlaced fashion. The time at which images are collected is determined by the

repetition time and the echo time. The repetition time is the time interval between

successive excitation pulses while the echo time is the time interval between an excitation

pulse and data acquisition. The RARE (Rapid Acquisition with Relaxation Enhancement)


sequence has evenly-spaced (echo spacing) multiple refocusing pulses to compensate for

the gradual loss of phase coherence after initial excitation forming an echo train (multiple

echoes). The echoes in each echo train are used to acquire multiple lines of data from each

signal excitation and reduce the overall time required to collect the image. T2-weighted,

coronal images were acquired using this RARE sequence with an echo spacing of 11 ms,

RARE factor (number of echoes in an echo train) of 8 and other parameters as in Table 2

below. In T2-weighted images, compartments filled with water like the cerebrospinal fluid

appear bright while tissues with a high lipid content like the white matter appear dark. Radio

frequency pulse shapes were calculated automatically using the Shinnar-Le Roux radio

frequency pulse design algorithm (Shinnar et al. 1989a, b, Shinnar et al. 1989c, Shinnar and

Leigh 1989, Pauly et al. 1991). Before running the fMRI imaging sequence, the homogeneity

of the local magnetic field was optimised in the region of interest using Bruker’s map shim

routine. After shimming, T2* fMRI images were acquired using a single-shot gradient-echo

echo planar imaging (EPI) sequence with the parameters in Table 2 below. Gradient-echo

EPI allows for the acquisition of a series of two-dimensional images by changing spatial

gradients rapidly following a single radio frequency pulse from a transmitter coil.

Table 2. MRI parameters used in the RARE anatomical scan and the functional T2* scan using single-shot gradient-echo echo planar imaging (EPI).

Parameters RARE T2* EPI

TR (repetition time) 2500 ms 1500 ms

TE (echo time) 33 ms 11 ms

Scan Time 2 min 55 s 7 min 30 s

FOV (field of view) 28.0 x 28.0 mm2 28.2 x 21.0 mm2

Matrix size 280 x 280 94 x 70

Resolution 0.1 mm x 0.1 mm 0.3 mm x 0.3 mm

Receiver Bandwidth 34722.2 Hz 326087.0 Hz

No of slices (coronal) 21 21

Slice Thickness 1 mm 1 mm

Slice gap 0.05 mm 0.05 mm

Read orientation Left to right Left to right


Slice ordering Non-contiguous, interlaced Non-contiguous, interlaced

Partial Fourier Acquisition N/A 0.829 (58/70)

Shots N/A 1

Dummy scans 2 8

Repetitions 1 300

NA (No of averages) 2 1

Flip angle (α) 900 900

Flip back pulse

Fat suppression

2.6 DATA ANALYSIS

Most of the analyses were carried out in the FSL v5.0 software package (Functional MRI of

the Brain (FMRIB) Software Library) (Jenkinson et al. 2012). Given that most data analysis

tools for fMRI, including FSL, are specifically adapted for handling human fMRI data, many

procedures were manually performed and additional steps were necessary for analysis of

rat brain as compared to human brains.

2.6.1 fMRI data processing

The MRI scanner outputs the neuroimaging data in DICOM (Digital Imaging and

Communications in Medicine) format, the international standard for medical images and

related information format (Bidgood Jr et al. 1997) (http://dicom.nema.org/). In order to be

viewed and analysed in FSL, the DICOM format data was first converted into NIFTI

(Neuroimaging Informatics Technology Initiative, https://nifti.nimh.nih.gov/) file format

using the dcm2nii converter command line tool (Appendix 1A) (Rorden et al. 2007). Pre-

processing of fMRI acquisitions included (i) upscaling the voxel sizes by a factor of 10 to be

closer to the size of a human brain (Tambalo et al. 2015), (ii) carrying out motion correction

using the FSL/MCFLIRT (FMRIB Linear Image Registration Tool with Motion Correction)

command (Jenkinson et al. 2002) to spatially realign the 6,300 functional images (21 slices

with 300 repetitions) to the middle volume of a serial acquisition to compensate for any

breathing-related movements during acquisition and (iii) reorienting the brain in LAS axes

(radiological view), which is the standard orientation in FSL (Appendix 1B).


To prevent extra brain matter from interfering with the results, it is recommended to

remove all non-brain structures (skull, muscles and surrounding tissue) from the data before

any co-registration can be performed in FSL. However, because this software was designed

primarily for human fMRI data, the FSL/BET tool (FMRIB Brain Extraction Tool) is not suitable

for rodent data. Instead, intracranial binary brain masks were created manually using ITK-

SNAP 3.4.0 (Yushkevich et al. 2006) (www.itksnap.org) for each functional and anatomical

data sets and were used to extract the brain using the flsmaths tool in the command line

(Appendix 1C). Post-stimulation images were co-registered to the baseline fMRI image

(Appendix 1D) using 6 parameter (three rotations and three translations in the x, y and z

planes) Rigid Body registration with the default correlation ratio cost metric in the FSL Linear

Image Registration Tool (FLIRT) (Jenkinson and Smith 2001, Jenkinson et al. 2002). Given

that the two data sets were acquired on the same day for the same animal, the size of the

brain in the images would be the same, which means there was no need to estimate x, y

and z scaling parameters during co-registration.

A single-session independent component analysis (ICA) was carried out for each brain-

extracted image in FSL/MELODIC (Multivariate Exploratory Linear Decomposition into

Independent Components) (Beckmann et al. 2005) to remove high- and low-frequency

noise from the data (Appendix 1E) (Salimi-Khorshidi et al. 2014). This process included in-

plane spatial smoothing using a Gaussian kernel filter set to the default full-width half

maximum (FWHM) of 5 mm to increase the signal-to-noise ratio by reducing small variances

and thermal noise. Pre- and post- 10 Hz stimulation data for the 6 animals were also

smoothed with FWHM set to 8 and 12 mm to investigate the effect of smoothing. The

default temporal high pass filter cut-off at 100 s was also applied to all data sets to reduce

hardware noise and low-frequency signal drifts.

ICA separates signals into independent components. Based on the characteristics (spatial,

temporal and frequency domains) of these components, they were then manually labelled

as ‘signal’ or ‘noise’ (Appendix 1F). On the threshold spatial maps, noise components are

located outside or at the edge of the brain (motion artefact) and display spatial overlap with

ventricles, brain blood vessels or brain stem (physiological noise) (Figure 16). Noise

http://www.itksnap.org/


components are predominantly high-frequency in the temporal power spectra and show

sudden spikes in the time series (Supplementary Figure 1). Signal components, on the other

hand, are principally characterised by low-frequency power spectra and form ‘blob-like’

activation patterns on spatial maps primarily covering the grey matter. Once all of the

components were classified, the data was ‘cleaned’ by removing the noise components

from the data using the fsl_regfilt command on the filtered data obtained from MELODIC

(Appendix 1F).

The pre- and post-stimulation de-noised fMRI images for each session were then co-

registered to their respective high-resolution structural T2-weighted images using 6

parameter Rigid Body registration (for the same reasons described above), which resulted

in realignment to the anatomical template (Lu et al. 2010). To normalise these realigned

images and facilitate automated processing, a Sprague Dawley brain atlas (Papp et al. 2014,

Kjonigsen et al. 2015, Sergejeva et al. 2015) was used as a standard template. The atlas was

first down-sampled by a factor of 8 (3.125 mm x 3.125 mm x 3.125 mm) to better match the

voxel size of the 4D functional data. Given that the size of the atlas brain was different from

the brain size of the rats used in this experiment, 9 degrees of freedom (6 parameters as

before, plus independent scaling in x, y and z directions) “traditional” registration was used

in FLIRT for co-registration of the images to the atlas (Appendix 1G). All subsequent analyses

were conducted in the atlas standard space.

2.6.2 fMRI statistical analyses

Multi-session/multi-subject temporal concatenation group ICA was performed to

determine between group differences by comparing pre- and post-stimulation fMRI images

and within group differences by performing the group ICA on all pre-stimulation data and

each post-stimulation data (sham, 1 Hz, 10 Hz, BHFS, cTBS) separately. Pre-processed

functional data for each animal were temporally concatenated across subjects to create a

single 4D dataset containing all data from for that particular group (e.g. all post- 10 Hz

stimulation datasets). The ICA algorithm was restricted to 15 components on the basis of

other rs-fMRI studies in rodents (Jonckers et al. 2011, Zerbi et al. 2015) and was performed

with the MELODIC toolbox (Appendix 2A). Given the limited sample size and the novelty of


the parameters of interest, we report the results based on a cluster-forming threshold of z

> 2, corresponding to an uncorrected p-value < 0.0455 for a two-tailed hypothesis as both

increases and decreases in activity were tested for. MELODIC outputs were IC maps

obtained from an alternative hypothesis test at p > 0.5, thresholded by fitting a mixture

model to the histogram of intensity values (Beckmann and Smith 2004).

Dual regression (Beckmann et al. 2009, Filippini et al. 2009) was then performed on the

resultant ICA components (Appendix 2B) to investigate group differences in resting-state

networks. Dual regression maps group-ICA results back into individual subjects’ data for

examination of between-group difference in ICA networks. These subject-specific spatial

maps were used to investigate intra- and inter-subject variability qualitatively (Section 3.2.2,

see Appendix 2E for commands run in the terminal). The design matrices and contrasts for

the dual regressions were set up in the FSL/GLM (General Linear Model) tool

(Supplementary Figure 2). Contrasts use the design matrices to specify the type of statistical

test (e.g. group mean, paired difference) the dual regression algorithm will carry out. Single

group paired differences were tested between pre- and post- stimulation rs-fMRI data by

conducting paired t-tests, and group averages for post-stimulation and pre-stimulation

groups were tested by conducting a one-sample t-test. The dual regression results are (1-p)

value maps and t-stat maps for each ICA component.

The group-ICA components for the pre-stimulation group (z > 2) were visually inspected,

and the DMN identified based on the spatial patterns in reference to known anatomical and

functional locations using a rat brain atlas (Paxinos and Watson 1998). After identifying the

DMN using group-level ICA on the rs-fMRI data, pre- and post-rTMS homologous ICA

components (z > 2) were visually compared to determine the effect of rTMS on the DMN.

The brain regions in the DMN showing a difference in correlation were identified using the

brain atlas (Supplementary Figure 4). ICA is a less biased approach relative to statistical

analyses such as seed-based connectivity analysis (Cole et al. 2010). Seed-based correlation

analysis finds the temporal correlation of all voxels within the brain based on a specified

seed voxel or small region of interest chosen beforehand (Joel et al. 2011). Seed-points can

differ in their location, which considerably affects the connectivity patterns and renders


between study comparisons unreliable (Jansen et al. 2015). In contrast, the use of data-

driven ICA may improve reproducibility because there is no (arbitrary) seed selection

(Rosazza et al. 2012). This approach enabled identification of brain regions that exhibit a

considerable difference in correlation between the pre- and post- rTMS groups without a

priori knowledge.


CHAPTER THREE: RESULTS

3.1 PHYSIOLOGICAL RECORDING

Rat body temperature was maintained at 37-38°C while under anaesthesia on a

temperature-controlled bed. The temperature of the bed was adjusted to keep the rat body

temperature within the specified range. The breathing rate, detected by the pneumatic

pillow, was maintained at 40-60 breaths/minute and the pulse at 250-500 beats/minute by

adjusting the concentration of isoflurane in medical oxygen. Oxygen saturation of the blood,

as detected by the pulse oximeter stayed in the range 95-100%. Physiological parameters

were stable (within the specified range) for all rats during the entire scan, and stimulation

period (Figure 13). No physical response was observed when the coil was turned on or off

during stimulation, and there were no abrupt changes in the physiological recordings for

any of the animals.

Figure 13. Representative physiological recordings for two animals demonstrating that body temperature (degrees Fahrenheit, A), respiratory rate (breaths per minute, B), oxygen saturation (%, C) and pulse (beats per minute, D) were stable. The upper and lower

limits of the y-axes have been set to the range within which each physiological recording should lie. Body temperature (A) was maintained between 980F (~370C) - 1010F (~380C), respiratory rate (B) between 40-60 breaths/min, oxygen saturation (C) between 95-100% and pulse between 250-500 bpm.


3.2 GROUP-ICA COMPONENTS

3.2.1 Default mode network in the pre-stimulation group

As noted, rs-fMRI was used to identify the DMN in rats in this study. The DMN is one of

many resting-state networks in the brain which is comprised of interacting brain regions

having a highly synchronised activity distinct from other networks in the brain. The group-

ICA algorithm was run on the pre-stimulation group, and the resting-state networks were

identified (see Figure 14 for a 3D-rendered image produced using the MRIcroGL software

(Rorden and Brett 2000)). The components obtained from MELODIC were overlaid on the

rat brain template to which they were originally co-registered (Section 2.6.1) and the

distribution of the synchronised voxels was investigated using the digital brain atlas labels.

Based on visual inspection of the spatial map for each of the 15 components and the

consistency of the spatial distribution with known neuroanatomical regions from the brain

atlas, six meaningful non-artefactual circuits could be identified, which formed part of the

putative DMN (Figure 15). The remaining nine components were classified as noise. They

comprised components with low z-scores and the clusters, if any, were outside the brain,

near the brain surface or in the brain stem (examples are shown in Figure 16).

The DMN in rats has previously been described as consisting of the orbital cortex (Lu et al.

2012, Zerbi et al. 2015, Sierakowiak et al. 2015, Huang et al. 2016, Hsu et al. 2016), cingulate

cortex (Jonckers et al. 2011, Lu et al. 2012, Zerbi et al. 2015, Sierakowiak et al. 2015, Huang

et al. 2016, Hsu et al. 2016), auditory cortex (Lu et al. 2012, Huang et al. 2016, Hsu et al.

2016), somatosensory cortex (Huang et al. 2016), striatum/caudate putamen (Huang et al.

2016), retrosplenial cortex (Lu et al. 2012, Zerbi et al. 2015, Sierakowiak et al. 2015, Huang

et al. 2016), temporal association cortex (Lu et al. 2012, Zerbi et al. 2015, Hsu et al. 2016),

prelimbic cortex (Zerbi et al. 2015, Sierakowiak et al. 2015, Hsu et al. 2016), parasubiculum

(Zerbi et al. 2015), entorhinal cortex (Zerbi et al. 2015), hippocampus (Lu et al. 2012, Huang

et al. 2016, Hsu et al. 2016) and visual cortex (Lu et al. 2012, Hsu et al. 2016). Figure 15

provides the ICA components with clusters corresponding to these regions. The motor

cortex and inferior colliculus, which form part of the resting-state network (Jonckers et al.

2011) have also been identified. Most of the chosen ICA components (Figure 15) had

spatially asymmetrical correlations between homologous brain regions. In some


components, the homologous brain region was completely absent (no correlation at that

particular time point) while in some components, the spatial extent of the clusters was

larger in one hemisphere. The ‘dominant’ hemisphere with increased ipsilateral cluster size

was the same between sessions and across all animals (Figure 17).

Figure 14. Independent component maps of pre-stimulation rs-fMRI group overlaid on 3D-rendered standard Sprague Dawley brain atlas. The figure shows the superior view (top left),

anterior view (top right) and lateral view (bottom) of independent components from the group-ICA. The spatial colour-coded z-maps of these components are overlaid on the brain atlas (down-sampled by a factor of 8). A higher z-score (bright red) represents a higher correlation between the time course of that voxel and the mean time course of this component. Colour bar indicates z-scores (n = 24, thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).


Figure 15. Thresholded independent component maps showing the default mode network in the pre-stimulation rs-fMRI group-ICA dataset. The figure shows six independent components

from the group-ICA, three coronal slices and three coronal slices with their corresponding axial slices. The spatial colour-coded z-maps of these components are overlaid on the brain atlas (down-sampled by a factor of 8). A higher z-score (yellow) represents a higher correlation between the time course of that voxel and the mean time course of this component. Significant clusters within the components include various brain regions: 1, orbital cortex; 2, cingulate cortex; 3, auditory cortex; 4, somatosensory cortex; 5, striatum/caudate putamen; 6, retrosplenial cortex; 7, temporal association cortex; 8, prelimbic cortex; 9, parasubiculum; 10, entorhinal cortex; 11, hippocampus; 12, visual cortex; 13, inferior colliculus; 14, motor cortex. R denotes right hemisphere. x, y, z refer to the coordinates in standard Sprague Dawley template space. Colour bar indicates z-scores (n = 24, thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).


Figure 16. Two independent components — one axial slice (A) and one coronal slice (B) — from group-ICA classified as ‘noise’ components. A shows highly synchronised resting activity in the

brainstem while B shows highly synchronised activity only at the dorsal part of the coronal slice. Colour bar indicates z-scores (thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).

3.2.2 Reproducibility over time and between subjects in the pre-stimulation group

Because we scanned each animal at four different timepoints, the reproducibility of the

group-ICA results over time and between subjects could be investigated. All the pre-

stimulation datasets were thresholded to a z-score higher than 2, binarised (i.e., z-score < 2

set to 0, and z-score > 2 set to 1) and then summed (Appendix 2E) to give cumulative

reproducibility maps for each subject (Figure 17). The regions where the voxels have a z-

score of 2 or higher for one, two, three or four sessions are shown in different colours. The

reproducibility maps illustrate that the middle part of the clusters show overlap for all or at

least three time points while the border of the clusters represents data from only one or

two time points. Most of the scatter, and single-voxel correlations come from single sessions

(shown in grey). This shows that the central resting-state activity is reproducible, even when

the rs-fMRI data is acquired after a week or more. The same pattern is seen in each of the

six animals, showing that the data is reproducible between subjects as well.


Figure 17. Reproducibility between sessions and between subjects of a representative pre-stimulation group-ICA component. The reproducibility maps are session cumulative score maps

over the different time points overlaid on the Sprague Dawley rat brain atlas (down-sampled by a factor of 8). x, y, z refer to the coordinates in standard Sprague Dawley template space. Colour code: voxels with z-value greater than 2 (uncorrected p-value < 0.0455 for a two-tailed hypothesis) for: 1 session, grey; 2 sessions, dark brown; 3 sessions, orange; 4 sessions, yellow.

3.2.3 Pre- vs. post-stimulation resting-state activity

3.2.3.1 Dual regression results

Dual regression performs a between-group analysis using subject-specific spatial maps and

a set of defined contrasts (Supplementary Figure 2). The outputs are (1 – p-value) maps such

that greater intensity values signify smaller p-values. These p-values are corrected for

multiple comparisons across voxels for each resting state network (each ICA map) for a one-

tailed t-test for the contrast specified (pre > post and post > pre, Appendix 2B). Given that

six out of 15 group-ICA components consisted of the DMN, Bonferroni (Bonferroni 1935)

corrected p-values were < 0.05 / (6*2) = 0.0042 (z-score of 2.86). The p-value images

showed no statistically significant differences between pre- and post- stimulation in the

resting-state networks. In practice, group-ICA results are typically not Bonferroni corrected

(e.g., Figure 7 in Calhoun et al. 2009, Supplementary Figure 1 in Jansen et al. 2015).

3.2.3.2 Comparison of DMN in ICA components

Group-ICA components for each of the post-stimulation datasets were compared to the

DMN components identified in the pre-stimulation group to investigate the effect of rTMS.


Homologous components were found, showing high correlation in the same brain regions

(Supplementary Table 3). There were no unilateral changes in any of the DMN components

after sham stimulation (Figure 18).

Figure 18. Homologous ICA components showing that there is no change in the default mode network in isoflurane-anaesthetized rats before and after sham stimulation. The post

sham stimulation colour-coded z-maps were derived from dual regressed group-ICA on 6 animals and were overlaid on the Sprague Dawley brain template (down-sampled by a factor of 8). Coronal and axial slices for two representative ICA components are shown before (left) and after (right) sham stimulation. A higher z-score (yellow) represents a higher correlation between the time course of that voxel and the mean time course of this component. R denotes right hemisphere. x, y, z refer to the coordinates in standard Sprague Dawley template space. Colour bar indicates z-scores (thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).

There were clear changes in the synchronised activity following active rTMS stimulation

(Figure 19 and Table 3). Both excitatory frequencies (10 Hz and BHFS) showed more

noticeable ipsilateral changes in the strength of correlation within the DMN (Figure 19). The

inhibitory frequency, 1 Hz, showed bilateral changes in most components, although there

were more contralateral changes in the synchronised activity of brain regions (Table 3). cTBS

caused more ipsilateral increases in synchrony, and its effects were not as widespread as

the other rTMS protocols (See Supplementary Figure 3). Only large changes in synchrony of

resting activity have been noted, whereby a whole brain region, as identified in the atlas

(Appendix 3), showed a change.


Figure 19. Homologous group-ICA components showing changes in the default mode network of isoflurane-anaesthetized rats before and after rTMS stimulation at four frequencies: 10 Hz, BHFS, 1 Hz and cTBS. The post-stimulation colour-coded z-maps were derived

from group-ICA on six animals and were overlaid on the Sprague Dawley brain template (down-sampled by a factor of 8). Coronal and axial slices for two representative ICA components are shown before (left) and after (right) stimulation at each of the four rTMS protocols. A higher z-score (yellow) represents a higher correlation between the time course of that voxel and the mean time course of this component. Significant clusters within the components include various brain regions: 1, orbital cortex; 2, cingulate cortex; 3, auditory cortex; 4, somatosensory cortex; 5, striatum/caudate putamen; 6, retrosplenial cortex; 7, temporal association cortex; 8, prelimbic cortex; 9, parasubiculum; 10, entorhinal cortex; 11, hippocampus; 12, visual cortex; 13, inferior colliculus; 14, motor cortex. R denotes right hemisphere. x, y, z refer to the coordinates in standard Sprague Dawley template space. Colour bar indicates z-scores (thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).


Table 3. Summary of changes in the default mode network in brain regions post-rTMS using 10 Hz and BHFS excitatory frequencies and 1 Hz and cTBS inhibitory frequencies on the right (R) side (ipsilateral changes) and on the left (L) side (contralateral changes) of the brain. The numbers in brackets correspond to the numbers used to denote the different brain regions

in Figures 15 and 20 above. “↑” refers to an increase in correlated activity and “↓” refers to a decrease in correlated activity.

Region Function

Excitatory frequencies

Inhibitory frequencies

10 Hz BHFS 1 Hz cTBS

R L R L R L R L

Orbital cortex (1) Decision making and emotional processing ↑ ↑ - - ↑ ↑ ↑ -

Auditory cortex (3)

Hearing functions ↓ - ↓ ↓ - ↓ - ↓

Somatosensory cortex* (4)

Receives all sensory input ↓ - ↓ ↓ ↓ ↓ ↑ -

Striatum (5) Facilitates voluntary movement ↓ - ↓ - ↓ - ↑ ↑

Retrosplenial cortex (6)

Episodic memory and spatial navigation ↓ - - - ↓ ↓ - -

Entorhinal cortex (10)

Learning and memory ↓ - ↓ - - - - -

Hippocampus* (11)

Regulates emotions, spatial learning and memory

↓ - ↓ ↓ ↓ ↓ - -

Visual cortex (12)

Receives and processes visual input - - - ↓ - ↓ - -

Inferior colliculus (13)

Part of the auditory system ↓ - ↓ - - - - -

Motor cortex* (14)

Motor function and motor planning ↓↑ ↑ - ↓ ↑ ↓ ↑ -

*These brain regions are the primary targets of stimulation.

10 Hz stimulation caused bilateral changes in the synchrony of resting activity in the orbital

cortex and motor cortex and an ipsilateral decrease in the auditory cortex, somatosensory


cortex, striatum, retrosplenial cortex, entorhinal cortex, hippocampus and inferior

colliculus.

BHFS caused a bilateral decrease in the synchrony of resting activity of the auditory cortex,

somatosensory cortex and hippocampus, an ipsilateral decrease in the synchrony of the

striatum, entorhinal cortex and inferior colliculus and a contralateral decrease in the visual

and motor cortex.

1 Hz stimulation caused bilateral changes in the synchrony of resting activity of the orbital

cortex, somatosensory cortex, retrosplenial cortex, hippocampus and motor cortex, a

contralateral decrease in the auditory cortex and visual cortex and an ipsilateral decrease

in the striatum.

cTBS caused a bilateral increase in the synchrony of resting activity of the striatum, an

ipsilateral increase in the orbital cortex, somatosensory cortex and motor cortex and a

contralateral decrease in the auditory cortex.

3.2.4 Effect of smoothing

To better clarify the influence of smoothing, we performed the single-subject MELODIC ICA

decomposition at three smoothing levels (5 mm, 8 mm and 12 mm) on the pre- and post-

10 Hz stimulation datasets. This step was necessary as this was the first time rs-fMRI has

been used to study the effects of rTMS in rodents. Hence, we explored the impact of

smoothing on our model to determine its effect on statistical significance and whether

smoothing would compromise our ability to distinguish differences in correlation in

different brain regions.

An increase in smoothing resulted in an increase in the total number of components

obtained from the single-session ICA but a decrease in the percentage of useful (not noise)

components (Table 4). When group-ICA was run on the pre- 10 Hz stimulation rs-fMRI

datasets for the six animals, the number of voxels with a high z-score (yellow in colour)

increased with increased smoothing (Figure 20A). Dual regression with post- minus pre-

stimulation contrast did not show any significant results when thresholded to p < 0.05,

meaning that the differences in resting-state activity between the pre- and post-stimulation

groups were not statistically significant. However, there appeared to be a decrease in the


‘spotty’ pattern of activation in the t-stat activation map (Figure 20B) with increased

smoothing. The 12 mm FWHM filter resulted in less ‘speckles’, but the clusters of activation

for t > 2 were still small enough to be able to find the corresponding brain regions.

Therefore, 12 mm smoothing, which has been used in previous rs-fMRI studies in rats

(Hutchison et al. 2010), appeared to work best for our study.

Table 4. Smoothing increased the total number of components and decreased the percentage of useful ICA components obtained from single-session MELODIC on the pre- and post- 10 Hz rs-fMRI datasets for the six rats. Spatial smoothing was carried out using a Gaussian

kernel filter set to the default full-width half maximum (FWHM) of 5 mm, 8 mm and 12 mm.

Smoothing Mean total number of

components

Mean percentage of useful

components

5 mm 16 20%

8 mm 38 19%

12 mm 133 15%


Figure 20. Effect of smoothing on individual ICA components for the pre-stimulation group (A) and the t-stat activation maps of post- minus pre- 10 Hz stimulation of a representative animal (B). Spatial smoothing was carried out using a Gaussian kernel filter set to the default full-width half

maximum (FWHM) of 5 mm, 8 mm and 12 mm for the pre- and post- 10 Hz stimulation data for the six animals. The spatial colour-coded z-maps (A) and t-maps (B) are overlaid on the brain atlas (down-sampled by a factor of 8). A higher z-score (yellow) represents a higher correlation between the time course of that voxel and the mean time course of this component. Colour bar indicates z-scores for A (n = 24, thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis) and t-scores for B (n = 6, thresholded at t > 2).


CHAPTER FOUR: DISCUSSION

In recent years, resting-state networks have been found to be compromised in several

neurological and psychiatric conditions (van den Heuvel and Hulshoff Pol 2010) and rTMS

has been shown to modulate the activity of these networks in humans (Fox et al. 2012). In

this study, we explored how different LI-rTMS protocols alter the DMN in rats as measured

during rest and under anaesthesia using an offline combination of LI-rTMS and rs-fMRI. We

compared spontaneous activity in the brain during rest before and after the rats received

active or sham LI-rTMS over the right brain hemisphere. The DMN in the rat brain was

inferred based on synchronous fluctuations of the haemodynamic signals identified by ICA

of pre-stimulation rs-fMRI data. ICA of post-stimulation rs-fMRI data showed that 10 Hz

stimulation, BHFS and cTBS caused mostly ipsilateral changes in synchrony of activity in the

DMN while 1 Hz stimulation resulted in bilateral changes in synchrony, with the

contralateral changes being more prominent than ipsilateral changes. These findings

suggest that, in rTMS treatment, the frequency used should be carefully chosen depending

on the desired change in the brain regions within the DMN and the hemisphere being

stimulated. One clear limitation of this study is a small sample size, which could be the

reason why the difference in resting state activity pre- and post-stimulation did not reach

statistical significance in the dual regression analysis. We also assessed the impact of choice

of statistical approach, data smoothing and anaesthetics in the interpretation of the results.

4.1 THE DEFAULT MODE NETWORK BEFORE STIMULATION

Using ICA, a data-driven fMRI analysis approach, the DMN in the anaesthetised rat brains

was identified in the pre-stimulation group. The 3D-rendered images of the ICA components

showed bilateral symmetry in resting-state activity (Figure 14). However, the ICA maps

showed a variable degree of synchrony with other structures and consequently, did not

display bilateral hemispheric synchrony in resting activity (Figure 15). Coherent neuronal

oscillations or spontaneous rhythmic activity are believed to show which brain regions are

coupled for joint processing for a specific function, and the resulting hemodynamic

responses are interpreted as functional connectivity between these areas (Nikouline et al.

2001). MacDonald et al. (1996) measured the oscillations of the auditory and


somatosensory cortex in anaesthetized rats and found that the oscillations in the

somatosensory cortex were stronger in one hemisphere. Other similar behavioural and

electrophysiological studies have shown that homologous brain regions can function both

unilaterally and bilaterally (Banich and Belger 1990, Nikouline et al. 2001). This functional

ability is a possible explanation for the apparently stronger synchrony in resting-state

activity unilaterally in some brain regions within some ICA components (Figure 15). For

example, the first ICA component in Figure 15 shows that there is a strong synchrony in the

resting activity of the right auditory cortex (3) and right striatum/caudate putamen (5), but

there are no clusters for homologous brain regions in the left hemisphere. Asymmetry in

functional networks has previously been reported in resting-state network studies using ICA

(Fransson et al. 2009, Hutchison et al. 2010, Zerbi et al. 2015). The unilateral components

could represent stronger local connectivity, which could be both independent of, and

synchronised with, the inter-hemispheric connectivity within the DMN.

Interestingly, some brain regions, including the auditory cortex (3) and striatum (5), show

unilateral synchrony in some components and bilateral synchrony in others. A previous

study using ICA to identify resting-state networks in rats also reported that functionally

connected regions can split into separate components (Hutchison et al. 2010). Similar

observations were made when the group-ICA algorithm was limited to a different number

of components (See Supplementary Table 2). Having more components causes resting-state

activity of brain regions, which are part of the same network, to split into different

components based on higher local synchrony in activity (Jonckers et al. 2011, Zerbi et al.

2015). Overall, these results confirm that the group-ICA algorithm can cause homologous

brain regions within the DMN to appear in separate components and the extent of this is

dependent on the strength of bilateral synchrony and the total number of components. The

stronger the bilateral synchrony, the less likely they are to split into separate components

and the greater the number of components, the more likely it is for homologous brain

regions to split into separate components.

4.2 PRE- VS. POST-STIMULATION RESTING-STATE ACTIVITY

After identifying the DMN using group-level ICA on the rs-fMRI data, pre- and post-LI-rTMS

homologous ICA components were compared to determine the effect of LI-rTMS on the


DMN. For the purpose of this experiment, we will focus our discussion on the effect of LI-

rTMS on the somatosensory cortex, motor cortex and hippocampus.

4.2.1 Sham stimulation does not affect the DMN

Because of methodological differences between human and rodent MRI protocols, a key

control experiment performed here was to deliver sham stimulation to the rats. In human

studies with interleaved rTMS/fMRI protocols, the subjects do not have to come out of the

MRI instrument to obtain the brain stimulation. However, in our study, after the baseline

fMRI images were acquired, the imaging cradle was removed from the MRI machine for the

animal to receive the stimulation. Sudden changes in the animal’s surroundings, for

example, a change in temperature due to the temperature difference inside and outside the

scanner could potentially induce a change in the animal’s physiology and neural activity.

Physical handling of the imaging cradle and animal during the removal of the surface coil

from the bed and positioning of the LI-rTMS coil on the rat’s scalp could also cause some

stress to the animal, even when under anaesthetic, and hence induce a change in brain

activity. Therefore, sham stimulation was performed on each animal (0 Hz) with the coil

switched off as a control to ensure that the changes observed in neural activity was not due

to a change in the animal’s environment. When the pre-stimulation group-ICA components

were compared to the post-sham stimulation group, no unilateral changes in the DMN were

found (Figure 18), thus giving us confidence that our approach to study the effect of LI-rTMS

using rs-fMRI would provide data that was relevant to the brain changes induced by LI-rTMS.

4.2.2 Active LI-rTMS disrupts the DMN in rat brain

A change in the synchrony of resting-state activity in a specific brain region can be related

to either increase or decrease in activity of that area compared to other brain regions within

the same network. A decrease in synchrony, for example, does not necessarily mean a

decrease in activity and thus, we have related the observed changes in synchrony to what

is known about excitability changes reported in the literature of rTMS. However, given that

human studies use rTMS intensities that are much stronger than those used here in LI-rTMS,

the comparisons that are drawn are based on the assumption that LI-rTMS has frequency-

specific effects that are similar to those of rTMS delivered at higher intensities.


4.2.2.1 Somatosensory and Motor cortex

High-frequency rTMS (≥ 5 Hz) is known to cause an increase in excitability while 1 Hz and

cTBS are known inhibitory frequencies, and therefore cause a decrease in excitability in

human brains (for review, see Klomjai et al. 2015). An fMRI study by Schneider et al. (2010)

showed that there was an increase in activity of the targeted brain region when 5 Hz rTMS

(considered to be excitatory and have roughly equivalent effects to 10 Hz, see Wilson and

St George 2016) was applied over the left primary somatosensory cortex. In the current

study, 10 Hz stimulation was used, and an ipsilateral decrease in the synchrony of the resting

activity of the somatosensory cortex was observed. Given that high-frequency rTMS

stimulation is known to be excitatory, it is anticipated that this change could be related to

an increase in activity in the ipsilateral somatosensory cortex induced by 10 Hz LI-rTMS.

In contrast, when comparing the spatial maps post-cTBS with the pre-stimulation group, an

increase in the synchronised activity in the ipsilateral compared to the contralateral

somatosensory cortex (Figure 19) could clearly be seen. This result is in agreement with a

study by Rai et al. (2012) evaluating the effect of cTBS over the left somatosensory cortex

by measuring the change in tactile acuity of the contralateral hand. They noted a decrease

in neural activity within the stimulated cortex after cTBS application, which may indicate

that cTBS can reduce sensory processing in the ipsilateral cortex. The increase in synchrony

of resting-state activity in our results and the localised decrease in activity in the ipsilateral

somatosensory cortex following cTBS in Rai’s study appear to reflect comparable trends,

which implies the potential of cTBS to affect the activity within the stimulated cortex.

Outcomes following 1 Hz LI-rTMS stimulation are less clear-cut. While previous reports

suggest a purely uni-hemispheric effect, ours show bilateral and even contralateral

outcomes. When Enomoto et al. (2001) examined the changes in excitability of the sensory

cortex following rTMS stimulation by measuring changes in the somatosensory evoked

potentials, they found that 1 Hz rTMS over the left M1 suppressed the activity of only the

ipsilateral sensory cortex. Similar effects of 1 Hz stimulation were obtained by Vidoni et al.

(2010) a few years later. They studied the impact of 1 Hz rTMS over left primary

somatosensory cortex by observing cutaneous somatosensation and found that the


ipsilateral left hand was not affected by rTMS, but cutaneous somatosensation was found

to be significantly impaired in the right hand. In other words, rTMS elicited no change in

excitability of the contralateral somatosensory cortex. In contrast to this work, we found

that the synchrony of the resting state activity of the somatosensory cortex was decreased

bilaterally following 1 Hz stimulation. Some components showed a contralateral decrease

in synchrony exclusively, while those with bilateral decrease a showed a stronger ipsilateral

decrease in the synchrony of resting activity compared to the contralateral decrease within

the same component.

Although our findings regarding the effect of 1 Hz rTMS on the somatosensory cortex were

not entirely consistent with the human studies discussed above, there is also evidence from

studies in other brain regions that 1 Hz rTMS has effects on the contralateral hemisphere,

matching some of our results on synchrony of resting activity. 1 Hz being an inhibitory

frequency is thought to decrease the activity of inhibitory neurones in the stimulated

hemisphere, causing a reduction in the inhibitory interhemispheric drive, which in turn leads

to an increase in excitability of the contralateral hemisphere. This effect of 1 Hz rTMS has

been exploited in treating stroke patients by applying low-frequency stimulation to the

unaffected hemisphere to decrease transcallosal inhibition of the lesioned hemisphere and

consequently improve motor function in stroke patients (Takeuchi et al. 2005, Mansur et al.

2005, Fregni et al. 2006). The fact that the motor cortex in both hemispheres experiences a

change in neuronal excitability following 1 Hz rTMS may explain the bilateral changes in

synchrony observed in the motor cortex in our study. Interestingly, applying high-frequency

rTMS to the lesioned hemisphere can have a similar effect by improving ipsilesional

hemispheric excitability and hence improving motor rehabilitation. In a stroke study, 5 Hz

(high-frequency) rTMS was applied ipsilesionally, and a bilateral increase in M1 connectivity

was found (Li et al. 2016). In accordance with this study, we also found a bilateral increase

in the synchrony of resting activity in the motor cortex following 10 Hz stimulation, a

frequency that has similar effects to 5 Hz (Wilson and St George 2016).

Previous human studies have found that there are bilateral changes in M1 activity following

cTBS stimulation (Huang et al. 2005, Huang and Mouraux 2015). However, in our results, a


more prominent ipsilateral increase in the synchronised activity of the motor cortex was

observed following cTBS LI-rTMS stimulation to the right hemisphere of the rat brains.

Although one can argue that there was a change contralaterally as well (Figure 19), the

degree of this change was not deemed substantial enough to be considered as an actual

change in the synchrony of resting-state activity. The intrinsic differences between the

methods used to detect changes in correlation and activity or the limitations in imaging

measurements like EPI distortions could be the cause of this inconsistency.

4.2.2.2 Hippocampus

The hippocampus is an important brain region involved in memory, learning and emotion.

It is located deep within the brain and therefore its activity is unlikely to be directly

modulated by LI-rTMS, which affects only superficial regions directly underneath the coil

(see Section 1.3). While the proximal changes in DMN may reflect the direct stimulation of

those brain regions, any change in the activity of the hippocampus would be indirect and

due to the modulation of functionally connected regions. We detected an ipsilateral

decrease in the synchronised activity of the hippocampus following 10 Hz LI-rTMS

stimulation. This result is supported by the associative memory study by Wang et al. (2014).

They applied high-frequency (20 Hz) stimulation to the left lateral parietal cortex of healthy

adults to non-invasively enhance the targeted cortical-hippocampal networks and study

their role in associative memory. An ipsilateral change in the hippocampus was detected

following multiple-session stimulation and the increased functional connectivity was

correlated with improved associative memory performance. Hence, our results displayed a

correlation profile that is coherent with what is known about the effect of high-frequency

rTMS on the hippocampus in the literature.

On the other hand, after 1 Hz LI-rTMS stimulation, we found a bilateral decrease in the

correlation of hippocampal activity relative to other brain regions. van der Werf et al. (2010)

also found that the hippocampus had reduced activation bilaterally following the

application of low-frequency rTMS over the left dorsolateral prefrontal cortex. They

hypothesised that this change was not due to direct stimulation because the changes in

neural activity were observed distally relative to the site of stimulation. Consistent with this


finding, the change in the synchronised activity of the hippocampus observed in the DMN

in our study could, therefore, be due to the change in cortical excitability or the transcallosal

spread of LI-rTMS effects inducing bilateral inhibition.

It is worth mentioning that BHFS is a relatively new pattern of stimulation and it has not yet

been used in human studies. As such, there is little information about its effects in the

literature. Two studies using BHFS LI-rTMS in mice have shown increased structural

plasticity of visual pathway topography in the midbrain, thalamus and cortex (Rodger et al.

2012, Makowiecki et al. 2014). In addition, an in vitro study identified intracellular calcium

increases and changes in gene expression in dissociated mouse cortical neurons after LI-

rTMS BHFS (Grehl et al. 2015). However, it remains unclear how these cellular and molecular

changes might relate to DMN changes. Like 1 Hz stimulation, BHFS also had a bilateral effect

on the synchronised activity in the somatosensory cortex and the hippocampus. The motor

cortex resting activity following BHFS LI-rTMS showed a completely different effect

compared to the other three rTMS protocols. There was a contralateral decrease in the

correlation of the motor cortex to other brain regions. Further studies in animals and

humans are warranted to further investigate the effects of BHFS on the motor cortex.

To date, all the studies on the effect of rTMS on the DMN’s structure and function have

been conducted in humans. To the best of our knowledge, the present study is the first to

show evidence of alterations in the DMN caused by LI-rTMS in a pre-clinical model and most

of these changes are supported by the literature. LI-rTMS is a relatively new approach and

the fact that it can elicit changes in the DMN is surprising given that no action potentials are

induced. Our findings provide a framework to further explore how LI-rTMS affects the

rodent DMN (using rs-fMRI) and its relevant functions (using behavioural tests) and

eventually to translate these protocols to humans.

4.3 DUAL REGRESSION: PRE- VS. POST- ACTIVE rTMS STIMULATION

Statistical analysis of rs-fMRI rodent data is very complex. Following the convention in this

field of brain imaging (Table 5), we reported z-scores rather than p-values. Even though

there were differences in the DMN between the pre- and post- active rTMS stimulation

(Figure 19), the dual regression results indicate that those changes were not statistically


significant. This issue could be related to the small sample size (n=6) in the present study,

but most people accept that the large variability means that the standard regression-type

approaches are not the most useful approach (Calhoun et al. 2009). Therefore, the results

presented are z-maps from the group-ICA, rather than p-value maps from the dual

regression analysis.

Table 5. Reported statistics in recent rs-fMRI rodent studies. Research Article Rodent Sample

Size Values reported

Huang et al. (2016) Rat 9 Correlation coefficients for p<0.0056, corrected

Sierakowiak et al. (2015)

Rat 20 Functional correlation values for p<0.0005, uncorrected

Tambalo et al. (2015) Rat 18+4 ctrl z > 5.6; (1-p) > 0.989 ≡ p < 0.011

Zerbi et al. (2015) Mouse 15 z > 6; z-score min=2.7, max=15

Lu et al. (2012) Rat 16 t>5.6, corrected p < 0.05

Jonckers et al. (2011) Rat Mouse

5 9

z > 0

Hutchison et al. (2010)

Rat 20 z > 0 For functional connectivity, z > 1

4.4 ASSESSMENT OF ANALYSIS PARAMETERS

4.4.1 Effect of smoothing

Resting-state correlation data is based on low-frequency signals, and therefore, spatial

smoothing is often used to improve signal-to-noise ratio without reducing valid activation.

Smoothing works by averaging data points with their neighbours in fMRI images within a

dataset independently (Mikl et al. 2008). For successful smoothing, the underlying

activation area should be larger than the extent of smoothing. In other words, for detecting

small areas of activation, lower smoothing, if any, should be used, but if larger areas of

activation are expected, smoothing can be increased. We observed that increasing

smoothing results in a reduction of ‘speckles’ and an increase in correlation (higher z-scores)

between the time course of the voxels within the large clusters and the mean time course

of the component (Figure 20A), indicating that increased smoothing is effective for

identifying major areas of activation. In the study by Tian et al. (2013) assessing the effects

of spatial and temporal features in ICA of rs-fMRI, they obtained similar results to our study

when comparing 10 mm smoothed spatial maps to 5 mm ones. They detected an increased

spatial smoothness and increased the number of significant voxels with higher spatial

smoothing.


Moreover, smoothing reduces random noise within fMRI images, allowing better detection

of real activations when performing statistical analyses. Data from other regions within the

brain is added to a particular brain region, resulting in more independent components

(Table 4). An increase in the number of components increases the number of independent

observations being averaged and hence improves the validity of statistical tests by making

the error distribution more normal (Tahmasebi et al. 2009). The percentage of useful

components decreases with increased smoothing, meaning that more ‘noise’ components

are identified and removed, making the resulting data ‘cleaner’.

Furthermore, without smoothing, there can be imperfect functional alignment between the

images of different subjects as activated brain regions are rarely represented in the same

voxels. Smoothing accommodates for these variations by increasing the overlap of activated

areas across subjects during co-registration of fMRI images of different animals to each

other in group-ICA to find spatial correspondence between the brains (Mikl et al. 2008). This

wider overlap between activated voxels explains the larger cluster size and reduction in

‘spotty’ pattern of activation in the post- minus pre- 10 Hz stimulation group (Figure 20B)

with increased smoothing.

The recommended spatial smoothing is usually at least twice the voxel size (Worsley and

Friston 1995). Nevertheless, in the literature, people use varying the amount of smoothing

(see examples in Supplementary Table 1). The voxel size of our fMRI images was 3 mm

(originally 0.3 mm but was upscaled by a factor of 10 during pre-processing, see section

2.6.1) and we utilised a value of 5 mm (0.5 mm for original data) smoothing on all the

datasets as most rodent studies use a smoothing of 0.3-0.6 mm. The drawbacks of high

spatial smoothing of fMRI data include decreased resolution, blurring and shifting of

clusters. For studies requiring detailed localisation of small clusters, limited spatial

smoothing should be carried out on the data while if large changes are expected, smoothing

may be used more generously. In short, the amount of smoothing for a particular study

should be chosen carefully depending on the resolution of activation needed. For the

purpose of our study, we could have used a higher smoothing value to increase the signal-

to-noise ratio and hence, obtain more statistically significant results.


4.4.2 Possible issues with rs-fMRI and rTMS in rodents

4.4.2.1 Maintaining stable experimental conditions

It is important to maintain stable conditions throughout an experiment and also between

sessions to avoid confounding variables. That no change in synchronised resting activity was

observed after sham stimulation (Figure 18) shows that the experimental conditions were

sufficiently stable. Moreover, the reproducibility maps (Figure 17) show that the correlation

at the centre of a cluster was always greater than a z-score of 2 irrespective of the timepoint

and animal. Therefore, the correlations observed do not depend on the timepoint or the

animal. However, several technical issues need to be considered when interpreting our

data.

Position of the animal during scan

The rats were manually placed on the MRI bed, and their heads immobilised using a bite bar

and ear bars. Although care had been taken to load each rat to the animal holder in a

consistent fashion, the positioning of the ear bars inside the ear canals could vary from

experiment to experiment, as could the positioning of the bite bar. Additionally, the

stimulation coil was positioned on the right side of the head, next to the ear but the exact

positioning may not have been consistent between experimental sessions (Section 2.4.1).

Maintaining stable physiological conditions

Fluctuations in physiological parameters can contribute to drifts in the BOLD signal causing

physiological artefacts in the fMRI data. These physiological artefacts can introduce a

significant and potentially confounding contribution to correlation measurements, thus

reducing the sensitivity to detect differences between pre- and post- rTMS groups. The

physiological parameters, including anaesthetic dose, body temperature, blood

oxygenation level, and respiratory and pulse rates were monitored and controlled (See

Section 2.3) during the entire experiment. Given that these parameters were stable over

the duration of the experiment (Figure 13) and comparable for the same animal on different

days and different animals, the effects of rTMS observed were not due to these basic

physiological processes (Section 3.1). Moreover, correct pre-processing and artefact-


cleaning procedure (Section 2.6.1) should decrease both intra- and inter-subject variability

in the rs-fMRI data (Zerbi et al. 2015). Low pass filtering prior to ICA removes most of these

effects, but low-frequency drifts may still remain and appear as independent components.

Signals from common sources, for example, due to breathing, are isolated into separate

components and therefore removed as noise from the single-session ICA data during pre-

processing. Additionally, fluctuations in BOLD signal caused by the cardiac cycle have been

shown to have a low contribution (about 1%) to the total variation in rs-fMRI data of rats

(Kalthoff et al. 2011). Therefore, the effect of the animals’ physiology on our fMRI data

would be minimal.

4.4.2.2 Use of anaesthetics

In human studies, it can be assumed that the physiological condition of the subject is

relatively constant throughout the rs-fMRI scan session (Pan et al. 2015). On the other hand,

in animal rs-fMRI, the use of anaesthesia is required to immobilise the animal and reduce

stress (Pan et al. 2015, Vincent et al. 2007). However, anaesthetics, including isoflurane

used in this experiment, may confound the imaging results as they may cause alterations in

neural activity, vascular reactivity and neurovascular coupling. These potential effects of

isoflurane should be considered when interpreting findings. Studies indicate that isoflurane

may affect the intracellular concentration of calcium (for review, see Gomez and

Guatimosim 2003), which may result in a change in presynaptic transmission or postsynaptic

excitability. Isoflurane decreases excitatory and increases inhibitory transmission, causing

an overall suppression of neural activity (Gomez and Guatimosim 2003, Ouyang and

Hemmings 2005). As such, in the presence of isoflurane, the ability of high-frequency rTMS

to depolarise is impaired (Gersner et al. 2011). Additionally, isoflurane, being a GABAergic

anaesthetic, induces vasodilation (Ohata et al. 1999), particularly in deep anaesthesia,

through the activation of ATP-sensitive potassium channels of smooth muscle cells in

cerebral arteries (Pan et al. 2015). Vasodilation leads to an increase in CBF, which may be

interpreted as an increase in activity. Despite these confounding factors, isoflurane is the

anaesthetic of choice for repeated long-term experiments (Masamoto and Kanno 2012).

The effect of injectable anaesthetics tends to wear off within 20 minutes and also vary

between individual animals, making it difficult to maintain a constant level of anaesthetic.


In contrast, isoflurane level can be kept within a specified range (1-2.5% in the present

study) within and between experiment sessions. The concentration of isoflurane can also

be adjusted (within the specified range: 1-2.5%) based on the physiological monitoring

recordings to keep the physiological parameters from fluctuating outside the set range

(Section 3.1). Moreover, the DMN has been shown to persist irrespective of the depth and

type of anaesthetics used (Lu et al. 2007, Vincent et al. 2007, Kannurpatti et al. 2008, Zhao

et al. 2008).

4.5 FUTURE DIRECTIONS

Further research into the refinement of rTMS protocols for animals and identifying the

mechanisms by which rTMS acts is required. Future studies should focus on the enormous

potential of different combinations of stimulation parameters such as optimal intensity,

duration of stimuli per session, the frequency of sessions and coil dynamics as they are

major determinants of the outcome of stimulation. To better understand the reported

clinical benefits of rTMS in different neurological conditions, fMRI technology can be used

to study the changes in brain activation in animal models specific to neurological disorders.

The use of such animal models will allow us to link the rTMS-induced changes in the DMN

(using rs-fMRI) to the changes in symptoms (through behavioural tests) following rTMS

treatment. There is also need to optimise an rs-fMRI protocol for rTMS rat models to

minimise artefacts and facilitate automated processing. More specifically, acquisition of a

magnetic field map for improved distortion correction of the EPI images will facilitate better

co-registration of images into a standard space. Similarly, collecting an EPI image with

swapped direction of data encoding can also help with correction of image distortions.

Instead of relying entirely on ICA clean-up to remove the effect of physiological noise, we

could also regress these confounding time series from the data in post-processing to obtain

a more significant result given that the physiological parameters were recorded by a

dedicated laptop and were time-locked to the rs-fMRI acquisition.

Diffusion tensor imaging (DTI) tractography can demonstrate the relationship between

neuronal fibre density and the functional connectivity within the DMN. Therefore, a DTI

study in parallel to an fMRI study could elucidate whether long-term treatment with rTMS


can also elicit changes in fibre tracts in the brain in the long run. Although there is a general

trend towards structural (DTI) and functional (rs-fMRI) connectivity being strongly

associated, it has been shown that there are some mismatches within the DMN whereby

regions showing high correlation have low fibre connectivity (Hsu et al. 2016). It will be

interesting to study whether the effect of rTMS on functional connectivity is related to the

fibre connectivity as well. Subsequent invasive techniques such as molecular studies can

then be used to explore those effects in greater detail and therefore provide information

about how observed functional changes in resting-state networks reflect the changes

detected at a molecular and cellular level.

4.6 CONCLUSION

This study presents preliminary evidence that demonstrates BOLD MRI signal changes after

LI-rTMS in rodents. We have shown that FWHM 12 mm smoothing increases the signal-to-

noise ratio and 15 components in group-ICA are sufficient to detect rTMS-induced changes

in the cortex, as well as in remote brain regions such as the hippocampus. Together, our

findings on the rat DMN underscore the ability of LI-rTMS to modulate resting-state activity

in interconnected brain regions. Hence, combined LI-rTMS-fMRI emerges as a powerful tool

to visualise rTMS-induced functional connectivity changes at a high spatio-temporal

resolution and help unravel the physiological processes causing the LI-rTMS-induced

changes observed in the cortex. Nonetheless, the precise mechanisms generating these

changes in the DMN remain to be elucidated. A more detailed understanding of the brain

structures targeted by LI-rTMS and an fMRI protocol optimised to detect those effects will

be essential for the interpretation of neuropsychological and cognitive effects of LI-rTMS in

a clinical context.


REFERENCES

Allen, Elena A., Brian N. Pasley, Thang Duong, and Ralph D. Freeman. 2007. "Transcranial magnetic stimulation elicits coupled neural and hemodynamic consequences." Science 317 (5846):1918-1921. doi: 10.1126/science.1146426.

Arias-Carrion, O. 2008. "Basic mechanisms of rTMS: Implications in Parkinson's disease." International Archives of Medicine 1 (1):2. doi: 10.1186/1755-7682-1-2.

Artola, Alain, and Wolf Singer. 1993. "Long-term depression of excitatory synaptic transmission and its relationship to long-term potentiation." Trends in Neurosciences 16 (11):480-487.

Banich, M. T., and A. Belger. 1990. "Interhemispheric interaction: How do the hemispheres divide and conquer a task?" Cortex 26 (1):77.

Barker, A. T., and I. Freeston. 2007. "Transcranial magnetic stimulation." Scholarpedia 2 (10):2936. doi: 10.4249/scholarpedia.2936.

Barker, A. T., R. Jalinous, and I. L. Freeston. 1985. "Non-invasive magnetic stimulation of human motor cortex." Lancet 1 (8437):1106.

Beck, Heinz, Ivan V. Goussakov, Ailing Lie, Christoph Helmstaedter, and Christian E. Elger. 2000. "Synaptic plasticity in the human dentate gyrus." Journal of Neuroscience 20 (18):7080-7086.

Beckmann, C. F., C. E. Mackay, N. Filippini, and S. M. Smith. 2009. "Group comparison of resting-state fMRI data using multi-subject ICA and dual regression." Neuroimage 47:S148-S148. doi: 10.1016/S1053-8119(09)71511-3.

Beckmann, C. F., and S. M. Smith. 2004. "Probabilistic independent component analysis for functional magnetic resonance imaging." IEEE Transactions on Medical Imaging 23 (2):137-152. doi: 10.1109/TMI.2003.822821.

Beckmann, Christian F., Marilena DeLuca, Joseph T. Devlin, and Stephen M. Smith. 2005. "Investigations into resting-state connectivity using independent component analysis." Philosophical transactions of the Royal Society of London. Series B, Biological sciences 360 (1457):1001-1013. doi: 10.1098/rstb.2005.1634.

Bestmann, S., J. Baudewig, H. R. Siebner, J. C. Rothwell, and J. Frahm. 2004. "Functional MRI of the immediate impact of transcranial magnetic stimulation on cortical and subcortical motor circuits." European Journal of Neuroscience 19 (7):1950-1962. doi: 10.1111/j.1460.9568.2004.03277.x.

Bestmann, Sven, Orlando Swayne, Felix Blankenburg, Christian C. Ruff, Patrick Haggard, Nikolaus Weiskopf, Oliver Josephs, Jon Driver, John C. Rothwell, and Nick S. Ward. 2008. "Dorsal premotor cortex exerts state-dependent causal influences on activity


in contralateral primary motor and dorsal premotor cortex." Cerebral Cortex 18 (6):1281-1291. doi: 10.1093/cercor/bhm159.

Bidgood Jr, W. Dean, Steven C. Horii, Fred W. Prior, and Donald E. Van Syckle. 1997. "Understanding and using DICOM, the data interchange standard for biomedical imaging." Journal of the American Medical Informatics Association 4 (3):199-212.

Biswal, B., F. Z. Yetkin, V. M. Haughton, and J. S. Hyde. 1995. "Functional connectivity in the motor cortex of resting human brain using echo-planar MRI." Magnetic resonance in medicine 34 (4):537.

Bliss, T. V. P., and T. Lomo. 1973. "Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path." The Journal of Physiology 232 (2):331-356. doi: 10.1113/jphysiol.1973.sp010273.

Bliss, Timothy V. P., and Sam F. Cooke. 2011. "Long-term potentiation and long-term depression: a clinical perspective." Clinics 66 (S1):3-17. doi: 10.1590/S1807-59322011001300002.

Bluhm, Robyn L., Jodi Miller, Ruth A. Lanius, Elizabeth A. Osuch, Kristine Boksman, R. W. J. Neufeld, Jean Théberge, Betsy Schaefer, and Peter Williamson. 2007. "Spontaneous low-frequency fluctuations in the BOLD signal in Schizophrenic patients: anomalies in the default network." Schizophrenia Bulletin 33 (4):1004-1012. doi: 10.1093/schbul/sbm052.

Bohning, Daryl E., Ananda Shastri, Ziad Nahas, Jeff P. Lorberbaum, Stig W. Andersen, Wayne R. Dannels, Ernst- U. Haxthausen, Diana J. Vincent, and Mark S. George. 1998. "Echoplanar BOLD fMRI of brain activation induced by concurrent transcranial magnetic stimulation." Investigative Radiology 33 (6):336-340. doi: 10.1097/00004424-199806000-00004.

Bohning, DarylE, Ananda Shastri, KathleenA McConnell, Ziad Nahas, JeffreyP Lorberbaum, DonnaR Roberts, Charlotte Teneback, DianaJ Vincent, and MarkS George. 1999. "A combined TMS/fMRI study of intensity-dependent TMS over motor cortex." Biological Psychiatry 45 (4):385-394. doi: 10.1016/S0006-3223(98)00368-0.

Bonferroni, C. E. 1935. "Il calcolo delle assicurazioni su gruppi di teste." In Studi in onore del Professore Salvatore ortu carboni, 13-60.

Bramham, Clive R. 2008. "Local protein synthesis, actin dynamics, and LTP consolidation." Current Opinion in Neurobiology 18 (5):524-531. doi: 10.1016/j.conb.2008.09.013.

Calhoun, Vince D., Jingyu Liu, and Tülay Adalı. 2009. "A review of group ICA for fMRI data and ICA for joint inference of imaging, genetic, and ERP data." Neuroimage 45 (1):S163-S172. doi: 10.1016/j.neuroimage.2008.10.057.


Caparelli, E. C., W. Backus, F. Telang, G. J. Wang, T. Maloney, R. Z. Goldstein, D. Anschel, and F. Henn. 2010. "Simultaneous TMS-fMRI of the visual cortex reveals functional network, even in absence of phosphene sensation." The open neuroimaging journal 4:100-110. doi: 10.2174/1874440001004010100.

Capone, F., M. Dileone, P. Profice, F. Pilato, G. Musumeci, G. Minicuci, F. Ranieri, R. Cadossi, S. Setti, P. A. Tonali, and V. Di Lazzaro. 2009. "Does exposure to extremely low frequency magnetic fields produce functional changes in human brain?" Journal of Neural Transmission 116 (3):257-265. doi: 10.1007/s00702-009-0184-2.

Cardenas-Morales, L., D. A. Nowak, T. Kammer, R. C. Wolf, and C. Schonfeldt-Lecuona. 2010. "Mechanisms and applications of theta-burst rTMS on the human motor cortex." Brain Topography 22 (4):294-306. doi: 10.1007/s10548-009-0084-7.

Chen, R., J. Classen, C. Gerloff, P. Celnik, E. M. Wassermann, M. Hallett, and Leonardo G. Cohen. 1997. "Depression of motor cortex excitability by low-frequency transcranial magnetic stimulation." Neurology 48 (5):1398-1403.

Cherkassky, Vladimir L., Rajesh K. Kana, Timothy A. Keller, and Marcel Adam Just. 2006. "Functional connectivity in a baseline resting-state network in autism." Neuroreport 17 (16):1687.

Clark, Caroline, Jeffrey Cole, Christine Winter, Kathy Williams, and Geoffrey Grammer. 2015. "A review of transcranial magnetic stimulation as a treatment for post-traumatic stress disorder." Current Psychiatry Reports 17 (10):1-9. doi: 10.1007/s11920-015-0621-x.

Cohen, D., and B. N. Cuffin. 1991. "Developing a more focal magnetic stimulation. 1. Some basic principles." Journal of Clinical Neurophysiology 8 (1):102-111.

Cole, David M., Stephen M. Smith, and Christian F. Beckmann. 2010. "Advances and pitfalls in the analysis and interpretation of resting-state fMRI data." Frontiers in Systems Neuroscience 4:8. doi: 10.3389/fnsys.2010.00008.

Corti, Manuela, Carolynn Patten, and William Triggs. 2012. "Repetitive transcranial magnetic stimulation of motor cortex after stroke: A focused review." American Journal of Physical Medicine and Rehabilitation 91 (3):254-270. doi: 10.1097/PHM.0b013e318228bf0c.

Davis, Simon W., Nancy A. Dennis, Norbou G. Buchler, Leonard E. White, David J. Madden, and Roberto Cabeza. 2009. "Assessing the effects of age on long white matter tracts using diffusion tensor tractography." NeuroImage 46 (2):530-541. doi: 10.1016/j.neuroimage.2009.01.068.

de Sauvage, R. C., I. Lagroye, B. Billaudel, and B. Veyret. 2008. "Evaluation of the potential genotoxic effects of rTMS on the rat brain and current density mapping." Clinical Neurophysiology 119 (2):482-91. doi: 10.1016/j.clinph.2007.09.137.


Dlabač-de Lange, Jozarni J., Rikus Knegtering, and André Aleman. 2010. "Repetitive transcranial magnetic stimulation for negative symptoms of schizophrenia: Review and meta-analysis." Journal of Clinical Psychiatry 71 (4):411-418. doi: 10.4088/JCP.08r04808yel.

Edwards, Mark J., Penelope Talleli, and John C. Rothwell. 2008. "Clinical applications of transcranial magnetic stimulation in patients with movement disorders." Lancet Neurology 7 (9):827-840. doi: 10.1016/S1474-4422(08)70190-X.

Enomoto, H., Y. Ugawa, R. Hanajima, K. Yuasa, H. Mochizuki, Y. Terao, Y. Shiio, T. Furubayashi, N. K. Iwata, and I. Kanazawa. 2001. "Decreased sensory cortical excitability after 1 Hz rTMS over the ipsilateral primary motor cortex." Clinical Neurophysiology 112 (11):2154-2158. doi: 10.1016/S1388-2457(01)00667-8.

Esslinger, Christine, Nadja Schüler, Carina Sauer, Dagmar Gass, Daniela Mier, Urs Braun, Elisabeth Ochs, Thomas G. Schulze, Marcella Rietschel, Peter Kirsch, and Andreas Meyer-Lindenberg. 2014. "Induction and quantification of prefrontal cortical network plasticity using 5 Hz rTMS and fMRI." Human Brain Mapping 35 (1):140-151. doi: 10.1002/hbm.22165.

Faraday, Michael, 1791-1897. 1965. Experimental researches in electricity: Dover Publications.

Filippini, Nicola, Bradley J. MacIntosh, Morgan G. Hough, Guy M. Goodwin, Giovanni B. Frisoni, Stephen M. Smith, Paul M. Matthews, Christian F. Beckmann, Clare E. Mackay, and Robert W. Mahley. 2009. "Distinct Patterns of Brain Activity in Young Carriers of the APOE-ε4 Allele." Proceedings of the National Academy of Sciences of the United States of America 106 (17):7209-7214. doi: 10.1073/pnas.0811879106.

Fitzgerald, Paul B., Kate Hoy, Susan McQueen, Jerome J. Maller, Sally Herring, Rebecca Segrave, Michael Bailey, Greg Been, Jayashri Kulkarni, and Zafiris J. Daskalakis. 2009. "A randomized trial of rTMS targeted with MRI based neuro-navigation in treatment-resistant depression." Neuropsychopharmacology 34 (5):1255-1262. doi: 10.1038/npp.2008.233.

Fox, M. D., M. A. Halko, M. C. Eldaief, and A. Pascual-Leone. 2012. "Measuring and manipulating brain connectivity with resting state functional connectivity magnetic resonance imaging (fcMRI) and transcranial magnetic stimulation (TMS)." Neuroimage 62 (4):2232-43. doi: 10.1016/j.neuroimage.2012.03.035.

Fox, P. T., M. E. Raichle, M. A. Mintun, and C. Dence. 1988. "Nonoxidative glucose consumption during focal physiologic neural activity." Science 241 (4864):462-464. doi: 10.1126/science.3260686.

Fox, Peter T., and Marcus E. Raichle. 1986. "Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human


subjects." Proceedings of the National Academy of Sciences of the United States of America 83 (4):1140-1144. doi: 10.1073/pnas.83.4.1140.

Fransson, Peter, Beatrice Skiold, Mathias Engstrom, Boubou Hallberg, Mikael Mosskin, Ulrika Aden, Hugo Lagercrantz, and Mats Blennow. 2009. "Spontaneous brain activity in the newborn brain during natural sleep: An fMRI study in infants born at full term." Pediatr Res 66 (3):301-305.

Fregni, Felipe, Paulo S. Boggio, Angela C. Valle, Renata R. Rocha, Julia Duarte, Merari J. L. Ferreira, Tim Wagner, Shirley Fecteau, Sergio P. Rigonatti, Marcelo Riberto, Steven D. Freedman, and Alvaro Pascual-Leone. 2006. "A sham-controlled trial of a 5-day course of repetitive transcranial magnetic stimulation of the unaffected hemisphere in stroke patients." Stroke 37 (8):2115-2122. doi: 10.1161/01.STR.0000231390.58967.6b.

Fujiki, Minoru, and Oswald Steward. 1997. "High frequency transcranial magnetic stimulation mimics the effects of ECS in upregulating astroglial gene expression in the murine CNS." Molecular Brain Research 44 (2):301-308. doi: 10.1016/S0169-328X(96)00232-X.

Funke, K., and A. Benali. 2011. "Modulation of cortical inhibition by rTMS - findings obtained from animal models." Journal of Physiology 589 (Pt 18):4423-35. doi: 10.1113/jphysiol.2011.206573.

Galhardoni, R., G. S. Correia, H. Araujo, L. T. Yeng, D. T. Fernandes, H. H. Kaziyama, M. A. Marcolin, D. Bouhassira, M. J. Teixeira, and D. C. de Andrade. 2015. "Repetitive transcranial magnetic stimulation in chronic pain: A review of the literature." Archives of Physical Medicine and Rehabilitation 96 (4):S156-S172. doi: 10.1016/j.apmr.2014.11.010.

Gaynes, B. N., S. W. Lloyd, L. Lux, G. Gartlehner, R. A. Hansen, S. Brode, D. E. Jonas, T. S. Evans, M. Viswanathan, and K. N. Lohr. 2014. "Repetitive transcranial magnetic stimulation for treatment-resistant depression: A systematic review and meta-analysis." Journal of Clinical Psychiatry 75 (5):477-U196. doi: 10.4088/JCP.13r08815.

George, M. S., and E. M. Wassermann. 1994. "Rapid-rate transcranial magnetic stimulation and ECT." Convulsive therapy 10 (4):251.

George, M. S., E. M. Wassermann, W. A. Williams, A. Callahan, T. A. Ketter, P. Basser, M. Hallett, and R. M. Post. 1995. "Daily repetitive transcranial magnetic stimulation (rTMS) improves mood in depression." Neuroreport 6 (14):1853.

Gersner, Roman, Elena Kravetz, Jodie Feil, Gaby Pell, and Abraham Zangen. 2011. "Long-term effects of repetitive transcranial magnetic stimulation on markers for neuroplasticity: differential outcomes in anesthetized and awake animals." The


Journal of neuroscience : the official journal of the Society for Neuroscience 31 (20):7521-7526. doi: 10.1523/JNEUROSCI.6751-10.2011.

Glielmi, Christopher B., Andrew J. Butler, Dmitriy M. Niyazov, Warren G. Darling, Charles M. Epstein, Jay L. Alberts, and Xiaoping P. Hu. 2014. "Assessing low-frequency repetitive transcranial magnetic stimulation with functional magnetic resonance imaging: A case series." Physiotherapy Research International 19 (2):117-125. doi: 10.1002/pri.518.

Gomez, Renato S., and Cristina Guatimosim. 2003. "Mechanism of action of volatile anesthetics: Involvement of intracellular calcium signaling." Current drug targets. CNS and neurological disorders 2 (2):123.

Gonzalez-Garcia, N., J. L. Armony, J. Soto, D. Trejo, M. A. Alegria, and R. Drucker-Colin. 2011. "Effects of rTMS on Parkinson's disease: A longitudinal fMRI study." Journal of Neurology 258 (7):1268-80. doi: 10.1007/s00415-011-5923-2.

Grehl, S., H. M. Viola, P. I. Fuller-Carter, K. W. Carter, S. A. Dunlop, L. C. Hool, R. M. Sherrard, and J. Rodger. 2015. "Cellular and molecular changes to cortical neurons following low intensity repetitive magnetic stimulation at different frequencies." Brain Stimulation 8 (1):114-123. doi: 10.1016/j.brs.2014.09.012.

Greicius, Michael D., Benjamin H. Flores, Vinod Menon, Gary H. Glover, Hugh B. Solvason, Heather Kenna, Allan L. Reiss, and Alan F. Schatzberg. 2007. "Resting-state functional connectivity in major depression: Abnormally increased contributions from subgenual cingulate cortex and thalamus." Biological Psychiatry 62 (5):429-437. doi: 10.1016/j.biopsych.2006.09.020.

Greicius, Michael D., Gaurav Srivastava, Allan L. Reiss, Vinod Menon, and Marcus E. Raichle. 2004. "Default-mode network activity distinguishes Alzheimer's disease from healthy aging: Evidence from functional MRI." Proceedings of the National Academy of Sciences of the United States of America 101 (13):4637-4642. doi: 10.1073/pnas.0308627101.

Hampson, M., and R. E. Hoffman. 2010. "Transcranial magnetic stimulation and connectivity mapping: Tools for studying the neural bases of brain disorders." Frontiers in Systems Neuroscience 4. doi: 10.3389/fnsys.2010.00040.

Hsu, L. M., X. Liang, H. Gu, J. K. Brynildsen, J. A. Stark, J. A. Ash, C. P. Lin, H. B. Lu, P. R. Rapp, E. A. Stein, and Y. H. Yang. 2016. "Constituents and functional implications of the rat default mode network." Proceedings of the National Academy of Sciences of the United States of America 113 (31):E4541-E4547. doi: 10.1073/pnas.1601485113.

Huang, E. J., and L. F. Reichardt. 2001. "Neurotrophins: Roles in neuronal development and function." Annual review of neuroscience 24:677.


Huang, Gan, and André Mouraux. 2015. "MEP latencies predict the neuromodulatory effect of cTBS delivered to the ipsilateral and contralateral sensorimotor cortex." PLoS One 10 (8). doi: 10.1371/journal.pone.0133893.

Huang, S. M., Y. L. Wu, S. L. Peng, H. H. Peng, T. Y. Huang, K. C. Ho, and F. N. Wang. 2016. "Inter-strain differences in default mode network: A resting state fMRI study on spontaneously hypertensive rat and Wistar Kyoto rat." Scientific Reports 6:21697. doi: 10.1038/srep21697.

Huang, Ying-Zu, Mark J. Edwards, Elisabeth Rounis, Kailash P. Bhatia, and John C. Rothwell. 2005. "Theta burst stimulation of the human motor cortex." Neuron 45 (2):201-206. doi: 10.1016/j.neuron.2004.12.033.

Hutchison, R. Matthew, Seyed M. Mirsattari, Craig K. Jones, Joseph S. Gati, and L. Stan Leung. 2010. "Functional networks in the anesthetized rat brain revealed by independent component analysis of resting-state fMRI." Journal of Neurophysiology 103 (6):3398-3406. doi: 10.1152/jn.00141.2010.

Iacoboni, Marco. 2008. "The role of premotor cortex in speech perception: Evidence from fMRI and rTMS." Journal of Physiology 102 (1–3):31-34. doi: 10.1016/j.jphysparis.2008.03.003.

Jaafari, Nematollah, Fady Rachid, Jean-Yves Rotge, Mircea Polosan, Wissam El-Hage, David Belin, Nicolas Vibert, and Antoine Pelissolo. 2012. "Safety and efficacy of repetitive transcranial magnetic stimulation in the treatment of obsessive-compulsive disorder: A review." World Journal of Biological Psychiatry 13 (3):164-177. doi: 10.3109/15622975.2011.575177.

Jansen, Jochem M., Guido van Wingen, Wim van den Brink, and Anna E. Goudriaan. 2015. "Resting state connectivity in alcohol dependent patients and the effect of repetitive transcranial magnetic stimulation." European Neuropsychopharmacology 25 (12):2230-2239. doi: 10.1016/j.euroneuro.2015.09.019.

Jenkinson, M., C. F. Beckmann, S. M. Smith, M. W. Woolrich, and T. E. J. Behrens. 2012. "FSL." NeuroImage 62 (2):782-790. doi: 10.1016/j.neuroimage.2011.09.015.

Jenkinson, Mark, Peter Bannister, Michael Brady, and Stephen Smith. 2002. "Improved optimization for the robust and accurate linear registration and motion correction of brain images." Neuroimage 17 (2):825-841. doi: 10.1006/nimg.2002.1132.

Jenkinson, Mark, and Stephen Smith. 2001. "A global optimisation method for robust affine registration of brain images." Medical Image Analysis 5 (2):143-156. doi: 10.1016/S1361-8415(01)00036-6.

Ji, Ru-Rong, Thomas E. Schlaepfer, Carlos D. Aizenman, Charles M. Epstein, Dike Qiu, Justin C. Huang, and Fabio Rupp. 1998. "Repetitive transcranial magnetic stimulation activates specific regions in rat brain." Proceedings of the National Academy of


Sciences of the United States of America 95 (26):15635-15640. doi: 10.1073/pnas.95.26.15635.

Joel, Suresh E., Brian S. Caffo, Peter C. M. van Zijl, and James J. Pekar. 2011. "On the relationship between seed-based and ICA-based measures of functional connectivity." Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 66 (3):644-657. doi: 10.1002/mrm.22818.

Jonckers, Elisabeth, Johan Van Audekerke, Geofrey De Visscher, Annemie Van der Linden, and Marleen Verhoye. 2011. "Functional connectivity fMRI of the rodent brain: comparison of functional connectivity networks in rat and mouse." PloS one 6 (4):e18876. doi: 10.1371/journal.pone.0018876.

Kähkönen, S., S. Komssi, J. Wilenius, and R. J. Ilmoniemi. 2005. "Prefrontal transcranial magnetic stimulation produces intensity-dependent EEG responses in humans." Neuroimage 24 (4):955-960. doi: 10.1016/j.neuroimage.2004.09.048.

Kalthoff, Daniel, Jörg U. Seehafer, Chrystelle Po, Dirk Wiedermann, and Mathias Hoehn. 2011. "Functional connectivity in the rat at 11.7T: Impact of physiological noise in resting state fMRI." NeuroImage 54 (4):2828-2839. doi: 10.1016/j.neuroimage.2010.10.053.

Kannurpatti, Sridhar S., Bharat B. Biswal, Young Ro Kim, and Bruce R. Rosen. 2008. "Spatio-temporal characteristics of low-frequency BOLD signal fluctuations in isoflurane-anesthetized rat brain." Neuroimage 40 (4):1738-1747. doi: 10.1016/j.neuroimage.2007.05.061.

Kelly, A. M. Clare, Lucina Q. Uddin, Bharat B. Biswal, F. Xavier Castellanos, and Michael P. Milham. 2008. "Competition between functional brain networks mediates behavioral variability." NeuroImage 39 (1):527-537. doi: 10.1016/j.neuroimage.2007.08.008.

Kennedy, Daniel P., Elizabeth Redcay, and Eric Courchesne. 2006. "Failing to deactivate: Resting functional abnormalities in autism." Proceedings of the National Academy of Sciences of the United States of America 103 (21):8275-8280. doi: 10.1073/pnas.0600674103.

Kimbrell, Timothy A., John T. Little, Robert T. Dunn, Mark A. Frye, Benjamin D. Greenberg, Eric M. Wassermann, Jennifer D. Repella, Aimee L. Danielson, Mark W. Willis, Brenda E. Benson, Andrew M. Speer, Elizabeth Osuch, Mark S. George, and Robert M. Post. 1999. "Frequency dependence of antidepressant response to left prefrontal repetitive transcranial magnetic stimulation (rTMS) as a function of baseline cerebral glucose metabolism." Biological Psychiatry 46 (12):1603-1613. doi: 10.1016/S0006-3223(99)00195-X.


Kjonigsen, Lisa J., Sveinung Lillehaug, Jan G. Bjaalie, Menno P. Witter, and Trygve B. Leergaard. 2015. "Waxholm Space atlas of the rat brain hippocampal region: Three-dimensional delineations based on magnetic resonance and diffusion tensor imaging." NeuroImage 108:441-449. doi: 10.1016/j.neuroimage.2014.12.080.

Klomjai, Wanalee, Rose Katz, and Alexandra Lackmy-Vallée. 2015. "Basic principles of transcranial magnetic stimulation (TMS) and repetitive TMS (rTMS)." Annals of Physical and Rehabilitation Medicine 58 (4):208-213. doi: 10.1016/j.rehab.2015.05.005.

Köhler, Stefan, Tomáš Paus, Randy L. Buckner, and Brenda Milner. 2004. "Effects of left inferior prefrontal stimulation on episodic memory formation: A two-stage fMRI—rTMS study." Journal of Cognitive Neuroscience 16 (2):178-188. doi: 10.1162/089892904322984490.

Kole, Maarten H. P., Eberhard Fuchs, Ulf Ziemann, Walter Paulus, and Ulrich Ebert. 1999. "Changes in 5-HT1A and NMDA binding sites by a single rapid transcranial magnetic stimulation procedure in rats." Brain Research 826 (2):309-312. doi: 10.1016/S0006-8993(99)01257-3.

Komssi, Soile, Seppo Kähkönen, and Risto J. Ilmoniemi. 2004. "The effect of stimulus intensity on brain responses evoked by transcranial magnetic stimulation." Human Brain Mapping 21 (3):154-164. doi: 10.1002/hbm.10159.

Korte, M, P Carroll, E Wolf, G Brem, H Thoenen, and T Bonhoeffer. 1995. "Hippocampal long-term potentiation is impaired in mice lacking brain-derived neurotrophic factor." Proceedings of the National Academy of Sciences 92 (19):8856-8860.

Lappchen, C. H., T. Ringer, J. Blessin, K. Schulz, G. Seidel, R. Lange, and F. Hamzei. 2015. "Daily iTBS worsens hand motor training - A combined TMS, fMRI and mirror training study." Neuroimage 107:257-65. doi: 10.1016/j.neuroimage.2014.12.022.

Larson, John, Darryl Wong, and Gary Lynch. 1986. "Patterned stimulation at the theta frequency is optimal for the induction of hippocampal long-term potentiation." Brain Research 368 (2):347-350. doi: 10.1016/0006-8993(86)90579-2.

Lee, Lucy, Hartwig Siebner, and Sven Bestmann. 2006. "Rapid modulation of distributed brain activity by transcranial magnetic stimulation of human motor cortex." Behavioural Neurology 17 (3-4):135-148. doi: 10.1155/2006/287276.

Lerman, Caryn, Hong Gu, James Loughead, Kosha Ruparel, Yihong Yang, and Elliot A. Stein. 2014. "Large-scale brain network coupling predicts acute nicotine abstinence effects on craving and cognitive function." JAMA Psychiatry 71 (5):523-530. doi: 10.1001/jamapsychiatry.2013.4091.

Li, Jing, Xue-Wei Zhang, Zhen-Tao Zuo, Jie Lu, Chun-Ling Meng, Hong-Ying Fang, Rong Xue, Yong Fan, Yu-Zhou Guan, and Wei-Hong Zhang. 2016. "Cerebral functional


reorganization in ischemic stroke after repetitive transcranial magnetic stimulation: An fMRI study." CNS Neuroscience & Therapeutics. doi: 10.1111/cns.12593.

Li, Xingbao, Karen J. Hartwell, Max Owens, Todd Lematty, Jeffrey J. Borckardt, Colleen A. Hanlon, Kathleen T. Brady, and Mark S. George. 2013. "Repetitive transcranial magnetic stimulation of the dorsolateral prefrontal cortex reduces nicotine cue craving." Biological psychiatry 73 (8):714-720. doi: 10.1016/j.biopsych.2013.01.003.

Lipton, Richard B., and Starr H. Pearlman. 2010. "Transcranial magnetic simulation in the treatment of migraine." Neurotherapeutics 7 (2):204-212. doi: 10.1016/j.nurt.2010.03.002.

Ljubisavljevic, Milos R., Asma Javid, Joji Oommen, Khatija Parekh, Nico Nagelkerke, Safa Shehab, and Thomas E. Adrian. 2015. "The effects of different repetitive transcranial magnetic stimulation (rTMS) protocols on cortical gene expression in a rat model of cerebral ischemic-reperfusion injury." PLoS One 10 (10). doi: 10.1371/journal.pone.0139892.

Lowe, Mark J., Micheal D. Phillips, Joseph T. Lurito, David Mattson, Mario Dzemidzic, and Vincent P. Mathews. 2002. "Multiple sclerosis: low-frequency temporal blood oxygen level-dependent fluctuations indicate reduced functional connectivity initial results." Radiology 224 (1):184-192. doi: 10.1148/radiol.2241011005.

Lu, Bai. 2003. "BDNF and activity-dependent synaptic modulation." Learning and memory 10 (2):86-98. doi: 10.1101/lm.54603.

Lu, Hanbing, Clara A. Scholl, Yantao Zuo, Steven Demny, William Rea, Elliot A. Stein, and Yihong Yang. 2010. "Registering and analyzing rat fMRI data in the stereotaxic framework by exploiting intrinsic anatomical features." Magnetic Resonance Imaging 28 (1):146-152. doi: 10.1016/j.mri.2009.05.019.

Lu, Hanbing, Qihong Zou, Hong Gu, Marcus E. Raichle, Elliot A. Stein, and Yihong Yang. 2012. "Rat brains also have a default mode network." Proceedings of the National Academy of Sciences of the United States of America 109 (10):3979-3984. doi: 10.1073/pnas.1200506109.

Lu, Hanbing, Yantao Zuo, Hong Gu, James A. Waltz, Wang Zhan, Clara A. Scholl, William Rea, Yihong Yang, and Elliot A. Stein. 2007. "Synchronized delta oscillations correlate with the resting-state functional MRI signal." Proceedings of the National Academy of Sciences of the United States of America 104 (46):18265-18269. doi: 10.1073/pnas.0705791104.

MacDonald, K. D., B. Brett, and D. S. Barth. 1996. "Inter- and intra-hemispheric spatiotemporal organization of spontaneous electrocortical oscillations." Journal of Neurophysiology 76 (1):423.


Machado, Sergio, Oscar Arias-Carrion, Fl Paes, Emmy Uehara, Bruna Velasques, Silmar Teixeira, Andre D. Santos, Claudio Queiroz, Jose Junior, Pedro Ribeiro, and Antonio E. Nardi. 2011. "Effects of repetitive transcranial magnetic stimulation on dystonia: An overview." American Journal of Neuroscience 2 (1):5-16. doi: 10.3844/amjnsp.2011.5.16.

Maeda, Fumiko, Julian P. Keenan, Jose M. Tormos, Helge Topka, and Alvaro Pascual-Leone. 2000. "Interindividual variability of the modulatory effects of repetitive transcranial magnetic stimulation on cortical excitability." Experimental Brain Research 133 (4):425-430. doi: 10.1007/s002210000432.

Makowiecki, K., A. R. Harvey, R. M. Sherrard, and J. Rodger. 2014. "Low-intensity repetitive transcranial magnetic stimulation improves abnormal visual cortical circuit topography and upregulates BDNF in mice." Journal of Neuroscience 34 (32):10780-92. doi: 10.1523/JNEUROSCI.0723-14.2014.

Malenka, Robert C., and Mark F. Bear. 2004. "LTP and LTD: An embarrassment of riches." Neuron 44 (1):5-21. doi: 10.1016/j.neuron.2004.09.012.

Mansur, C. G., F. Fregni, P. S. Boggio, M. Riberto, J. Gallucci-Neto, C. M. Santos, T. Wagner, S. P. Rigonatti, M. A. Marcolin, and A. Pascual-Leone. 2005. "A sham stimulation-controlled trial of rTMS of the unaffected hemisphere in stroke patients." Neurology 64 (10):1802.

Marcotte, Karine, Vincent Perlbarg, Guillaume Marrelec, Habib Benali, and Ana Inés Ansaldo. 2013. "Default-mode network functional connectivity in aphasia: therapy-induced neuroplasticity." Brain and language 124 (1):45-55. doi: 10.1016/j.bandl.2012.11.004.

Martiny, Klaus, Marianne Lunde, and Per Bech. 2010. "Transcranial low voltage pulsed electromagnetic fields in patients with treatment-resistant depression." Biological Psychiatry 68 (2):163-169. doi: 10.1016/j.biopsych.2010.02.017.

Masamoto, Kazuto, and Iwao Kanno. 2012. "Anesthesia and the quantitative evaluation of neurovascular coupling." Journal of Cerebral Blood Flow & Metabolism 32 (7):1233-1247. doi: 10.1038/jcbfm.2012.50.

McRobbie, Donald W., Elizabeth A. Moore, Martin J. Graves, and Martin R. Prince. 2007. MRI: from picture to proton. Vol. 2nd. Cambridge: Cambridge University Press.

Michael, Nikolaus, Michael Gösling, Michael Reutemann, Annette Kersting, Walter Heindel, Volker Arolt, and Bettina Pfleiderer. 2003. "Metabolic changes after repetitive transcranial magnetic stimulation (rTMS) of the left prefrontal cortex: a sham-controlled proton magnetic resonance spectroscopy (1H MRS) study of healthy brain." The European journal of neuroscience 17 (11):2462-2468. doi: 10.1046/j.1460-9568.2003.02683.x.


Mikl, Michal, Radek Mareček, Petr Hluštík, Martina Pavlicová, Aleš Drastich, Pavel Chlebus, Milan Brázdil, and Petr Krupa. 2008. "Effects of spatial smoothing on fMRI group inferences." Magnetic Resonance Imaging 26 (4):490-503. doi: 10.1016/j.mri.2007.08.006.

Miniussi, Carlo, and Paolo Maria Rossini. 2011. "Transcranial magnetic stimulation in cognitive rehabilitation." Neuropsychological Rehabilitation 21 (5):579-23. doi: 10.1080/09602011.2011.562689.

Minoshima, S., B. Giordani, S. Berent, K. A. Frey, N. L. Foster, and D. E. Kuhl. 1997. "Metabolic reduction in the posterior cingulate cortex in very early Alzheimer's disease." Annals of neurology 42 (1):85-94. doi: 10.1002/ana.410420114.

Mintun, Mark A., Brian N. Lundstrom, Abraham Z. Snyder, Andrei G. Vlassenko, Gordon L. Shulman, and Marcus E. Raichle. 2001. "Blood flow and oxygen delivery to human brain during functional activity: Theoretical modeling and experimental data." Proceedings of the National Academy of Sciences of the United States of America 98 (12):6859-6864. doi: 10.1073/pnas.111164398.

Morellini, N., S. Grehl, A. Tang, J. Rodger, J. Mariani, A. M. Lohof, and R. M. Sherrard. 2015. "What does low-intensity rTMS do to the cerebellum?" The Cerebellum 14 (1):23-26. doi: 10.1007/s12311-014-0617-9.

Muller-Dahlhaus, F., and A. Vlachos. 2013. "Unraveling the cellular and molecular mechanisms of repetitive magnetic stimulation." Frontiers in Molecular Neuroscience 6:50. doi: 10.3389/fnmol.2013.00050.

Müller, M. B., N. Toschi, A. E. Kresse, A. Post, and M. E. Keck. 2000. "Long-term repetitive transcranial magnetic stimulation increases the expression of brain-derived neurotrophic factor and cholecystokinin mRNA, but not neuropeptide tyrosine mRNA in specific areas of rat brain." Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology 23 (2):205-215. doi: 10.1016/S0893-133X(00)00099-3.

Muller, P. A., S. C. Dhamne, A. M. Vahabzadeh-Hagh, A. Pascual-Leone, F. E. Jensen, and A. Rotenberg. 2014. "Suppression of motor cortical excitability in anesthetized rats by low frequency repetitive transcranial magnetic stimulation." PLoS One 9 (3):e91065. doi: 10.1371/journal.pone.0091065.

Nettekoven, Charlotte, Lukas J. Volz, Martha Kutscha, Eva-Maria Pool, Anne K. Rehme, Simon B. Eickhoff, Gereon R. Fink, and Christian Grefkes. 2014. "Dose-dependent effects of theta burst rTMS on cortical excitability and resting-state connectivity of the human motor system." The Journal of neuroscience : the official journal of the Society for Neuroscience 34 (20):6849-6859. doi: 10.1523/JNEUROSCI.4993-13.2014.


Nikouline, V. V., K. Linkenkaer-Hansen, J. Huttunen, and R. J. Ilmoniemi. 2001. "Interhemispheric phase synchrony and amplitude correlation of spontaneous beta oscillations in human subjects: A magnetoencephalographic study." Neuroreport 12 (11):2487.

O’Reardon, John P., H. Brent Solvason, Philip G. Janicak, Shirlene Sampson, Keith E. Isenberg, Ziad Nahas, William M. McDonald, David Avery, Paul B. Fitzgerald, Colleen Loo, Mark A. Demitrack, Mark S. George, and Harold A. Sackeim. 2007. "Efficacy and safety of transcranial magnetic stimulation in the acute treatment of major depression: A multisite randomized controlled trial." Biological Psychiatry 62 (11):1208-1216. doi: 10.1016/j.biopsych.2007.01.018.

Oberman, Lindsay. 2014. "Repetitive transcranial magnetic stimulation (rTMS) protocols." Neuromethods 89:129-139. doi: 10.1007/978-1-4939-0879-0_7.

Ogawa, S., T. M. Lee, A. R. Kay, and D. W. Tank. 1990. "Brain magnetic resonance imaging with contrast dependent on blood oxygenation." Proceedings of the National Academy of Sciences of the United States of America 87 (24):9868-9872. doi: 10.1073/pnas.87.24.9868.

Ohata, Hiroto, Hiroki Iida, Shuji Dohi, and Yukinaga Watanabe. 1999. "Intravenous dexmedetomidine inhibits cerebrovascular dilation induced by isoflurane and sevoflurane in dogs." Anesthesia & Analgesia 89 (2):370-377. doi: 10.1097/00000539-199908000-00023.

Ouyang, Wei, and Hugh C. Hemmings. 2005. "Depression by isoflurane of the action potential and underlying voltage-gated ion currents in isolated rat neurohypophysial nerve terminals." Journal of Pharmacology and Experimental Therapeutics 312 (2):801-808. doi: 10.1124/jpet.104.074609.

Pan, Wen-Ju, Jacob C. W. Billings, Joshua K. Grooms, Sadia Shakil, and Shella D. Keilholz. 2015. "Considerations for resting state functional MRI and functional connectivity studies in rodents." Frontiers in neuroscience 9:269. doi: 10.3389/fnins.2015.00269.

Papp, Eszter A., Trygve B. Leergaard, Evan Calabrese, G. Allan Johnson, and Jan G. Bjaalie. 2014. "Waxholm Space atlas of the Sprague Dawley rat brain." NeuroImage 97:374-386. doi: 10.1016/j.neuroimage.2014.04.001.

Pascual-Leone, Alvaro, Josep Valls-Solé, Eric M. Wassermann, and Mark Hallett. 1994. "Responses to rapid-rate transcranial magnetic stimulation of the human motor cortex." Brain 117 (4):847-858. doi: 10.1093/brain/117.4.847.

Patterson, Susan L., Ted Abel, Thomas A. S. Deuel, Kelsey C. Martin, Jack C. Rose, and Eric R. Kandel. 1996. "Recombinant BDNF rescues deficits in basal synaptic transmission and hippocampal LTP in BDNF knockout mice." Neuron 16 (6):1137-1145. doi: 10.1016/S0896-6273(00)80140-3.


Pauly, J., P. Le Roux, D. Nishimura, and A. Macovski. 1991. "Parameter relations for the Shinnar-Le Roux selective excitation pulse design algorithm [NMR imaging]." IEEE Transactions on Medical Imaging 10 (1):53-65. doi: 10.1109/42.75611.

Paxinos, George, and Charles Watson. 1998. The rat brain in stereotaxic coordinates. 4th ed. London: Academic Press, Ltd.

Pell, G. S., Y. Roth, and A. Zangen. 2011. "Modulation of cortical excitability induced by repetitive transcranial magnetic stimulation: influence of timing and geometrical parameters and underlying mechanisms." Progress in Neurobiology 93 (1):59-98. doi: 10.1016/j.pneurobio.2010.10.003.

Pereira, Luisa Santos, Vanessa Teixeira Müller, Marleide da Mota Gomes, Alexander Rotenberg, and Felipe Fregni. 2016. "Safety of repetitive transcranial magnetic stimulation in patients with epilepsy: A systematic review." Epilepsy & Behavior 57 (Pt A):167-176. doi: 10.1016/j.yebeh.2016.01.015.

Popa, T., M. Russo, M. Vidailhet, E. Roze, S. Lehéricy, C. Bonnet, E. Apartis, A. P. Legrand, L. Marais, S. Meunier, and C. Gallea. 2013. "Cerebellar rTMS stimulation may induce prolonged clinical benefits in essential tremor, and subjacent changes in functional connectivity: An open label trial." Brain Stimulation 6 (2):175-179. doi: 10.1016/j.brs.2012.04.009.

Powers, W. J., I. B. Hirsch, and P. E. Cryer. 1996. "Effect of stepped hypoglycemia on regional cerebral blood flow response to physiological brain activation." American Journal of Physiology 270 (2):H554-H559.

Pridmore, Saxby, and Robert Belmaker. 1999. "Transcranial magnetic stimulation in the treatment of psychiatric disorders." Psychiatry and Clinical Neurosciences 53 (5):541-548. doi: 10.1046/j.1440-1819.1999.00603.x.

Rai, N., A. Premji, M. Tommerdahl, and A. J. Nelson. 2012. "Continuous theta-burst rTMS over primary somatosensory cortex modulates tactile perception on the hand." Clinical Neurophysiology 123 (6):1226-1233. doi: 10.1016/j.clinph.2011.09.026.

Raichle, M. E. 2009. "A brief history of human brain mapping." Trends in Neurosciences 32 (2):118-26. doi: 10.1016/j.tins.2008.11.001.

Raichle, Marcus E., Ann Mary MacLeod, Abraham Z. Snyder, William J. Powers, Debra A. Gusnard, and Gordon L. Shulman. 2001. "A default mode of brain function." Proceedings of the National Academy of Sciences of the United States of America 98 (2):676-682. doi: 10.1073/pnas.98.2.676.

Reithler, J., J. C. Peters, and A. T. Sack. 2011. "Multimodal transcranial magnetic stimulation: Using concurrent neuroimaging to reveal the neural network dynamics of noninvasive brain stimulation." Progress in Neurobiology 94 (2):149-165. doi: 10.1016/j.pneurobio.2011.04.004.


Ridding, M. C., and U. Ziemann. 2010. "Determinants of the induction of cortical plasticity by non‐invasive brain stimulation in healthy subjects." The Journal of Physiology 588 (13):2291-2304. doi: 10.1113/jphysiol.2010.190314.

Rodger, Jennifer, Christina Mo, Tenelle Wilks, Sarah A. Dunlop, and Rachel M. Sherrard. 2012. "Transcranial pulsed magnetic field stimulation facilitates reorganization of abnormal neural circuits and corrects behavioral deficits without disrupting normal connectivity." FASEB Journal 26 (4):1593-1606. doi: 10.1096/fj.11-194878.

Rodger, Jennifer, and Rachel M. Sherrard. 2015. "Optimising repetitive transcranial magnetic stimulation for neural circuit repair following traumatic brain injury." Neural Regeneration Research 10 (3):357-359. doi: 10.4103/1673-5374.153676.

Rohan, M. L., R. T. Yamamoto, C. T. Ravichandran, K. R. Cayetano, O. G. Morales, D. P. Olson, G. Vitaliano, S. M. Paul, and B. M. Cohen. 2014. "Rapid mood-elevating effects of low field magnetic stimulation in depression." BIOLOGICAL PSYCHIATRY 76 (3):186-193. doi: 10.1016/j.biopsych.2013.10.024.

Rorden, Chris, and Matthew Brett. 2000. "Stereotaxic display of brain lesions." Behavioural Neurology 12 (4):191-200. doi: 10.1155/2000/421719.

Rorden, Chris, Hans-Otto Karnath, and Leonardo Bonilha. 2007. "Improving lesion-symptom mapping." Journal of cognitive neuroscience 19 (7):1081-1088. doi: 10.1162/jocn.2007.19.7.1081.

Rosazza, C., L. Minati, F. Ghielmetti, M. L. Mandelli, and M. G. Bruzzone. 2012. "Functional connectivity during resting-state functional MR imaging: Study of the correspondence between independent component analysis and region-of-interest-based methods." AJNR. American journal of neuroradiology 33 (1):180. doi: 10.3174/ajnr.A2733.

Rossini, P. M., A. T. Barker, A. Berardelli, M. D. Caramia, G. Caruso, R. Q. Cracco, M. R. Dimitrijević, M. Hallett, Y. Katayama, and C. H. Lücking. 1994. "Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee." Electroencephalography and clinical neurophysiology 91 (2):79.

Rowland, Laura H., Robert S. Astur, Rex E. Jung, Juan R. Bustillo, John Lauriello, and Ronald A. Yeo. 2005. "Selective cognitive impairments associated with NMDA receptor blockade in humans." Neuropsychopharmacology 30 (3):633-639. doi: 10.1038/sj.npp.1300642.

Sale, Martin V., Michael C. Ridding, and Michael A. Nordstrom. 2007. "Factors influencing the magnitude and reproducibility of corticomotor excitability changes induced by paired associative stimulation." Experimental Brain Research 181 (4):615-626. doi: 10.1007/s00221-007-0960-x.


Sale, Martin V., Michael C. Ridding, and Michael A. Nordstrom. 2008. "Cortisol inhibits neuroplasticity induction in human motor cortex." Journal of Neuroscience 28 (33):8285-8293. doi: 10.1523/JNEUROSCI.1963-08.2008.

Salimi-Khorshidi, Gholamreza, Ludovica Griffanti, Junqian Xu, Clare E. Mackay, Claire E. Sexton, Karla L. Miller, Edward J. Auerbach, Kamil Ugurbil, Nicola Filippini, Klaus P. Ebmeier, Essa Yacoub, Steen Moeller, Steven M. Smith, Gwenaelle Douaud, Christian F. Beckmann, Giuseppe Baselli, and Eniko Zsoldos. 2014. "ICA-based artefact removal and accelerated fMRI acquisition for improved resting state network imaging." NeuroImage 95:232-247.

Sbardella, Emilia, Nikolaos Petsas, Francesca Tona, and Patrizia Pantano. 2015. "Resting-state fMRI in MS: General concepts and brief overview of its application." BioMed Research International 2015:1-8. doi: 10.1155/2015/212693.

Schneider, Susanne A., Burkhard Pleger, Bogdan Draganski, Carla Cordivari, John C. Rothwell, Kailash P. Bhatia, and Ray J. Dolan. 2010. "Modulatory effects of 5Hz rTMS over the primary somatosensory cortex in focal dystonia—An fMRI-TMS study." Movement Disorders 25 (1):76-83. doi: 10.1002/mds.22825.

Sergejeva, Marina, Eszter A. Papp, Rembrandt Bakker, Manuel A. Gaudnek, Yuko Okamura-Oho, Jyl Boline, Jan G. Bjaalie, and Andreas Hess. 2015. "Anatomical landmarks for registration of experimental image data to volumetric rodent brain atlasing templates." Journal of Neuroscience Methods 240:161-169. doi: 10.1016/j.jneumeth.2014.11.005.

Shinnar, M., L. Bolinger, and J. S. Leigh. 1989a. "The synthesis of soft pulses with a specified frequency-response." Magnetic Resonance in Medicine 12 (1):88-92. doi: 10.1002/mrm.1910120111.

Shinnar, M., L. Bolinger, and J. S. Leigh. 1989b. "The use of finite impulse-response filters in pulse design." Magnetic Resonance in Medicine 12 (1):81-87. doi: 10.1002/mrm.1910120110.

Shinnar, M., S. Eleff, H. Subramanian, and J. S. Leigh. 1989c. "The synthesis of pulse sequences yielding arbitrary magnetization vectors." Magnetic Resonance in Medicine 12 (1):74-80. doi: 10.1002/mrm.1910120109.

Shinnar, M., and J. S. Leigh. 1989. "The application of spinors to pulse synthesis and analysis." Magnetic Resonance in Medicine 12 (1):93-98. doi: 10.1002/mrm.1910120112.

Sierakowiak, Adam, Cyril Monnot, Sahar Nikkhou Aski, Martin Uppman, Tie-Qiang Li, Peter Damberg, and Stefan Brené. 2015. "Default mode network, motor network, dorsal and ventral basal ganglia networks in the rat brain: Comparison to human networks using resting state-fMRI." PLoS One 10 (3). doi: 10.1371/journal.pone.0120345.


Soleimani, Robabeh, Mir Mohammad Jalali, and Tolou Hasandokht. 2015. "Therapeutic impact of repetitive transcranial magnetic stimulation (rTMS) on tinnitus: A systematic review and meta-analysis." European Archives of Oto-Rhino-Laryngology. doi: 10.1007/s00405-015-3642-5.

Speer, Andrew M., Timothy A. Kimbrell, Eric M. Wassermann, Jennifer D. Repella, Mark W. Willis, Peter Herscovitch, and Robert M. Post. 2000. "Opposite effects of high and low frequency rTMS on regional brain activity in depressed patients." Biological Psychiatry 48 (12):1133-1141. doi: 10.1016/S0006-3223(00)01065-9.

Sutherland, Matthew T., Meredith J. McHugh, Vani Pariyadath, and Elliot A. Stein. 2012. "Resting state functional connectivity in addiction: Lessons learned and a road ahead." NeuroImage 62 (4):2281-2295. doi: 10.1016/j.neuroimage.2012.01.117.

Tahmasebi, Amir M., Purang Abolmaesumi, Zane Z. Zheng, Kevin G. Munhall, and Ingrid S. Johnsrude. 2009. "Reducing inter-subject anatomical variation: Effect of normalization method on sensitivity of functional magnetic resonance imaging data analysis in auditory cortex and the superior temporal region." Neuroimage 47 (4):1522-1531. doi: 10.1016/j.neuroimage.2009.05.047.

Takeuchi, Naoyuki, Takayo Chuma, Yuichiro Matsuo, Ichiro Watanabe, and Katsunori Ikoma. 2005. "Repetitive transcranial magnetic stimulation of contralesional primary motor cortex improves hand function after stroke." Stroke 36 (12):2681-2686. doi: 10.1161/01.STR.0000189658.51972.34.

Tambalo, Stefano, Luca Peruzzotti-Jametti, Roberta Rigolio, Silvia Fiorini, Pietro Bontempi, Giulia Mallucci, Beatrice Balzarotti, Paola Marmiroli, Andrea Sbarbati, Guido Cavaletti, Stefano Pluchino, and Pasquina Marzola. 2015. "Functional magnetic resonance imaging of rats with experimental autoimmune encephalomyelitis reveals brain cortex remodeling." The Journal of neuroscience 35 (27):10088.

Thickbroom, Gary W. 2007. "Transcranial magnetic stimulation and synaptic plasticity: experimental framework and human models." Experimental Brain Research 180 (4):583-593. doi: 10.1007/s00221-007-0991-3.

Tian, Lixia, Yazhuo Kong, Juejing Ren, Gaël Varoquaux, Yufeng Zang, and Stephen M. Smith. 2013. "Spatial vs. temporal features in ICA of resting-state fMRI - A quantitative and qualitative investigation in the context of response inhibition." PloS one 8 (6):e66572. doi: 10.1371/journal.pone.0066572.

van den Heuvel, Martijn P., and Hilleke E. Hulshoff Pol. 2010. "Exploring the brain network: A review on resting-state fMRI functional connectivity." European Neuropsychopharmacology 20 (8):519-534. doi: 10.1016/j.euroneuro.2010.03.008.

van der Werf, Ysbrand D., Ernesto J. Sanz-Arigita, Sanne Menning, and Odile A. van den Heuvel. 2010. "Modulating spontaneous brain activity using repetitive transcranial


magnetic stimulation." BMC Neuroscience 11 (1):145. doi: 10.1186/1471-2202-11-145.

van Praag, Henriette. 2009. "Exercise and the brain: something to chew on." Trends in Neurosciences 32 (5):283-290. doi: 10.1016/j.tins.2008.12.007.

Vercammen, Ans, Henderikus Knegtering, Edith J. Liemburg, Johannes A. den Boer, and André Aleman. 2010. "Functional connectivity of the temporo-parietal region in schizophrenia: Effects of rTMS treatment of auditory hallucinations." Journal of Psychiatric Research 44 (11):725-731. doi: 10.1016/j.jpsychires.2009.12.011.

Vidoni, E. D., N. E. Acerra, E. Dao, S. K. Meehan, and L. A. Boyd. 2010. "Role of the primary somatosensory cortex in motor learning: An rTMS study." Neurobiology of Learning and Memory 93 (4):532-539. doi: 10.1016/j.nlm.2010.01.011.

Vincent, J. L., G. H. Patel, M. D. Fox, A. Z. Snyder, J. T. Baker, D. C. Van Essen, J. M. Zempel, L. H. Snyder, M. Corbetta, and M. E. Raichle. 2007. "Intrinsic functional architecture in the anaesthetized monkey brain." Nature 447:83+. doi: 10.1038/nature05758.

Vlachos, Andreas, Florian Müller-Dahlhaus, Johannes Rosskopp, Maximilian Lenz, Ulf Ziemann, and Thomas Deller. 2012. "Repetitive magnetic stimulation induces functional and structural plasticity of excitatory postsynapses in mouse organotypic hippocampal slice cultures." Journal of Neuroscience 32 (48):17514-17523. doi: 10.1523/JNEUROSCI.0409-12.2012.

Waites, Anthony B., Regula S. Briellmann, Michael M. Saling, David F. Abbott, and Graeme D. Jackson. 2006. "Functional connectivity networks are disrupted in left temporal lobe epilepsy." Annals of neurology 59 (2):335-343. doi: 10.1002/ana.20733.

Walsh, Vincent. 1998. "Brain mapping: Faradization of the mind." Current Biology 8 (1):R8-R11. doi: 10.1016/S0960-9822(98)70098-3.

Wang, Fei, Xin Geng, Hua-Ying Tao, and Yan Cheng. 2010. "The restoration after repetitive transcranial magnetic stimulation treatment on cognitive ability of vascular dementia rats and its impacts on synaptic plasticity in hippocampal CA1 area." Journal of molecular neuroscience 41 (1):145-155. doi: 10.1007/s12031-009-9311-7.

Wang, Jane X., Lynn M. Rogers, Evan Z. Gross, Anthony J. Ryals, Mehmet E. Dokucu, Kelly L. Brandstatt, Molly S. Hermiller, and Joel L. Voss. 2014. "Targeted enhancement of cortical-hippocampal brain networks and associative memory." Science 345 (6200):1054-1057. doi: 10.1126/science.1252900.

Wassermann, Eric M., and Sarah H. Lisanby. 2001. "Therapeutic application of repetitive transcranial magnetic stimulation: a review." Clinical Neurophysiology 112 (8):1367-1377. doi: 10.1016/S1388-2457(01)00585-5.


Wassermann, Eric M., and Trelawny Zimmermann. 2012. "Transcranial magnetic brain stimulation: Therapeutic promises and scientific gaps." Pharmacology and Therapeutics 133 (1):98-107. doi: 10.1016/j.pharmthera.2011.09.003.

Waterhouse, Emily G., and Baoji Xu. 2009. "New insights into the role of brain-derived neurotrophic factor in synaptic plasticity." Molecular and Cellular Neuroscience 42 (2):81-89. doi: 10.1016/j.mcn.2009.06.009.

Whitfield-Gabrieli, Susan, Heidi W. Thermenos, Snezana Milanovic, Ming T. Tsuang, Stephen V. Faraone, Robert W. McCarley, Martha E. Shenton, Alan I. Green, Alfonso Nieto-Castanon, Peter LaViolette, Joanne Wojcik, D. E. Gabrieli John, Larry J. Seidman, and Marcus E. Raichle. 2009. "Hyperactivity and hyperconnectivity of the default network in Schizophrenia and in first-degree relatives of persons with Schizophrenia." Proceedings of the National Academy of Sciences of the United States of America 106 (4):1279-1284. doi: 10.1073/pnas.0809141106.

Wilson, Marcus T., and Lynley St George. 2016. "Repetitive transcranial magnetic stimulation: A call for better data." Frontiers in Neural Circuits 10 (57). doi: 10.3389/fncir.2016.00057.

Worsley, K. J., and K. J. Friston. 1995. "Analysis of fMRI time-series revisited—Again." Neuroimage 2 (3):173-181. doi: 10.1006/nimg.1995.1023.

Xia, Guohua, Prashant Gajwani, David J. Muzina, David E. Kemp, Keming Gao, Stephen J. Ganocy, and Joseph R. Calabrese. 2008. "Treatment-emergent mania in unipolar and bipolar depression: Focus on repetitive transcranial magnetic stimulation." The International Journal of Neuropsychopharmacology 11 (1):119-130. doi: 10.1017/S1461145707007699.

Yushkevich, Paul A., Joseph Piven, Heather Cody Hazlett, Rachel Gimpel Smith, Sean Ho, James C. Gee, and Guido Gerig. 2006. "User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability." Neuroimage 31 (3):1116-1128. doi: 10.1016/j.neuroimage.2006.01.015.

Zerbi, Valerio, Joanes Grandjean, Markus Rudin, and Nicole Wenderoth. 2015. "Mapping the mouse brain with rs-fMRI: An optimized pipeline for functional network identification." NeuroImage 123:11-21. doi: 10.1016/j.neuroimage.2015.07.090.

Zhang. 2007. "Effect of transcranial magnetic stimulation on the expression of c-Fos and brain-derived neurotrophic factor of the cerebral cortex in rats with cerebral infarct." Journal of Huazhong University Science, Technology and Medical Science 27 (4):415.

Zhao, Fuqiang, Tiejun Zhao, Lei Zhou, Qiulin Wu, and Xiaoping Hu. 2008. "BOLD study of stimulation-induced neural activity and resting-state connectivity in medetomidine-sedated rat." Neuroimage 39 (1):248-260. doi: 10.1016/j.neuroimage.2007.07.063.


APPENDICES

APPENDIX 1: fMRI IMAGE PRE-PROCESSING PROTOCOL

All data analyses were completed on an HP Z420 workstation with 64 GB of RAM and Linux

operating system CentOS 7.2.

A. Dicom to Nifti convertor

In terminal: dcm2niix -z y <insert target file>

B. X10 + motion correction + image reorientation

For functional images:

cp 2016*.nii.gz rs.nii.gz fslorient -setsform -0.3 0 0 0 0 0 1.05 0 0 -0.3 0 0 0 0 0 1 rs.nii.gz fslorient -copysform2qform rs.nii.gz fslswapdim rs.nii.gz x z -y rs.nii.gz cp rs.nii.gz rs_x10.nii.gz fslchpixdim rs_x10.nii.gz 3 10.5 3 mcflirt -in rs_x10.nii.gz

For T2 Anatomical:

cp 2016*.nii.gz t2.nii.gz fslorient -setsform -0.1 0 0 0 0 0 1.05 0 0 -0.1 0 0 0 0 0 1 t2.nii.gz fslorient -copysform2qform t2.nii.gz fslswapdim t2.nii.gz x z -y t2.nii.gz cp t2.nii.gz t2_x10.nii.gz fslchpixdim t2_x10.nii.gz 1 10.5 1

C. Brain extraction

ITK_SNAP: Manually create a mask using baseline resting state fMRI image and save as a

NIFTI file.

In terminal to swap dimensions because output from ITK_SNAP is flipped in the x

direction: fslswapdim <insert mask> -x y z <insert name of fixed mask>

In terminal for brain extraction: fslmaths <insert target image> -mas <insert fixed mask>

<insert name of new file with ‘_brain’>

D. Registration of post-stimulation fMRI image to baseline image


In terminal for registration using FLIRT: /usr/local/fsl/bin/flirt -in <insert post-stimulation

image from b> -ref <insert baseline image from b as reference image> -out <insert name of

output co-registered image as rs2_rs1.nii.gz> -omat <insert name of output co-registration

matrix as rs2_rs1.mat> -bins 35 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -

90 90 -dof 6 -interp trilinear

To correct volume: applyxfm4D <insert post-stimulation image from b> <insert baseline

image from b> <insert name of output image as rs2_rs1_4D.nii.gz> rs2_rs1.mat -

singlematrix

To correct repetition time (TR): cp rs2_rs1_4D.nii.gz rs2_rs1_4D_tr.nii.gz

fslchpixdim rs2_rs1_4D_tr.nii.gz 3 10.5 3 1.5

E. Single-session independent component analysis using MELODIC GUI

Data Tab

Input: Brain extracted images (rs1_brain.nii.gz for pre-stimulation and rs2_rs1_4D.nii.gz for

post-stimulation)

Output: The same directory

Pre-stats Tab

Motion Correction was switched off because it has already been carried out on the data.

BET brain extraction was deselected because the input images have been skull-stripped.

Spatial smoothing was left at the default value of 5 mm (also 8 and 12 mm for 10 Hz group).

Temporal filtering was left at the default setting (Highpass).

Registration Tab

No registrations were carried out at this stage.

Stats Tab

Single-Session ICA was selected.

All other settings were kept as default.


Supplementary Table 1: Examples of recent fMRI studies in rodents and humans and the respective fMRI data voxel size and smoothing used in pre-processing.

Study Subject Voxel Size

(mm)

Smoothing (mm) Smoothing factor

(to nearest 0.5)

(Zerbi et al. 2015) Mice 0.263 x

0.233

0.3 x 0.3 X 1-1.5

(Jonckers et al. 2011) Mice 0.16 x 0.31

x 0.4

0.3 x 0.4 X 1

Rats 0.23 x 0.23

x 1

0.4 x 0.4 X 2

(Hsu et al. 2016) Rats - 0.6 -

(Huang et al. 2016) Rats - 1 -

(Tambalo et al. 2015) Rats - 3 (0.3mm original

voxel size)

-

(Lu et al. 2012) Rats - 0.6 -

(Hutchison et al. 2010) Rats 0.4 1.2 X 3

(Sierakowiak et al. 2015) Rats - 0.6 -

Humans 3 8 X 2.5

(Lappchen et al. 2015) Humans 3 12 X 4

(Jansen et al. 2015) Humans - 5 -

(Esslinger et al. 2014) Humans 3 9 X 3

(Nettekoven et al. 2014) Humans 3.1 8 X 2.5

(Gonzalez-Garcia et al.

2011)

Humans 2 8 X 4

(Caparelli et al. 2010) Humans 3.125 mm 8 X 2.5

(van der Werf et al. 2010) Humans 2.3 6 X 2.5

(Vercammen et al. 2010) Humans - 10 -

(Bestmann et al. 2008) Humans 2 8 X 4

(Bestmann et al. 2004) Humans 2 3 X 1.5


F. De-noising

Supplementary Figure 1. Characteristic power spectra and time series for signal (A & C) and noise (B & D) components from single-session ICA. Signal components usually have a single

low-frequency peak in their power spectra (A) and no spikes in their time series (C) while noise components have multiple high-frequency peaks in their power spectra (B) and spikes in their time series (D).

The command below was run in a terminal with the working directory being the folder

created by MELODIC and a,b,c being the component numbers labelled as ‘noise’:

fsl_regfilt -i filtered_func_data.nii.gz -d filtered_func_data.ica/melodic_mix -f a,b,c -o

filtered_func_data_clean.nii.gz


G. Co-registration of denoised fMRI images to anatomical image and atlas

Due to the high resolution of the atlas, co-registrations of the low-resolution functional 4D

image data to the high-resolution anatomical image to the atlas crashed (Error message:

“Out-of-Memory”). Therefore, the atlas was down-sampled by a factor of 8 by running the

following command three times:

fslmaths <atlas> -subsamp2 <atlas_downsampled>

FLIRT GUI was used for the co-registrations using the down-sampled atlas. The functional

image was co-registered to the T2 anatomical image (high-resolution structural image)

using 6 degrees of freedom that was co-registered to the downsampled brain atlas

(standard space) using 9 degrees of freedom. The command lines output were as follows.

Registration of anatomical image to atlas

/usr/local/fsl/bin/flirt -in < t2_x10_brain.nii.gz > -ref <insert atlas> -omat rs_t2_atlas1.mat

-bins 35 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 9

Registration of functional image to anatomical image

/usr/local/fsl/bin/flirt -in filtered_func_data_clean.nii.gz -ref t2_x10_brain.nii.gz -omat

rs_t2_atlas2.mat -bins 35 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -

dof 6

Concatenation of the two transformation matrices obtained

/usr/local/fsl/bin/convert_xfm -concat rs_t2_atlas1.mat -omat rs_t2_atlas.mat

rs_t2_atlas2.mat

Registration of functional image to the atlas using concatenated matrix

/usr/local/fsl/bin/flirt -in filtered_func_data_clean.nii.gz -ref <insert atlas> -out

rs_t2_atlas.nii.gz -applyxfm -init rs_t2_atlas.mat -interp trilinear


APPENDIX 2: DATA ANALYSIS

A. Multi-session/subject ICA using MELODIC command line

The multi-session temporal concatenation analysis was carried out on each group

separately using the MELODIC command line in order to limit the number of components

to 15:

melodic -i inputlist.txt -o groupmelodic.ica -v --nobet --bgthreshold=10 --tr=1.500 --report

-d 15 --mmthresh=0.5 --Ostats -a concat

Group-ICAs with 30 components were also carried out on the post-10 Hz stimulation

group to determine the effect of the number of components (Supplementary Table 2). The

observed z-score ICA maps of 30-component analysis were very similar to the 15-

component analysis. However, the increased number of components caused components

belonging to the same functional networks to split into different components.

Supplementary Table 2. Comparison of group-ICA components when limiting the total number of components to 15 vs. 30 on the post- 10 Hz stimulation group.

Corresponding components

15 components 30 components

1 2, 7, 13

2 1

3 3

4 6, 28

5 4, 10, 17

6 5

7 9, 19, 21

8 8, 12, 20, 23

9 14, 27

10 17, 30

11 16, 25

12 15

13 17, 18, 29

14 24, 30

15 11, 26


Supplementary Table 3. Homologous ICA components (0 to 14) showing the DMN in the pre-stimulation group and the five post-stimulation groups (0 Hz, 10 Hz, BHFS, 1Hz and cTBS). Pre 0 Hz 10 Hz BHFS 1 Hz cTBS

0 3 0,5,3 2 4 0

1 0 2 0 3 1

2 1 1 3 2 4

3 - - - 0 2

4 2 - 1 1 3

11 11 10 10 9 10


B. Design and contrast matrices in the GLM tool

Supplementary Figure 2. Design and contrast matrices in FSL/GLM tool. Single group paired

difference was determined between pre- and post-stimulation by conducting a paired t-test (A) and group average for post-stimulation (B) and pre-stimulation (C) groups was determined by conducting a one-sample t-test.


C. Dual Regression

Command line:

dual_regression groupmelodic.ica/melodic_IC 1 design.mat design.con 5000

groupmelodic.ica/dual_regression `cat inputlist.txt`

Outputs obtained:

1) For each dataset, the selected set of spatial maps from the group-ICA was regressed

(as spatial regressors in a multiple regression) into the subject's 4D space-time

dataset to generate subject-specific time series.

2) Next, those time series were regressed (as temporal regressors) into the same 4D

data set, resulting in a set of subject-specific versions of the spatial maps, one per

group-level spatial map.

3) Then, within-group differences (pre- vs. post-stimulation) were tested for using FSL's

randomise permutation-testing tool. 5000 permutations were run (recommended

in FSL tutorials online)


D. Group-ICA comparison

The group-ICA components were 3D rendered in MRIcroGL software (Rorden and Brett

2000) using the ICA components as an overlay on the brain atlas.

Supplementary Figure 3. Independent component maps of rs-fMRI group-ICA datasets overlaid on 3D-rendered standard Sprague Dawley brain atlas showing differences in resting activity in pre-stimulation, 10 Hz, BHFS, 1 Hz and cTBS groups. The spatial colour-coded

z-maps of the group-ICA components are overlaid on the brain atlas (down-sampled by a factor of 8). A higher z-score (bright red) represents a higher correlation between the time course of that voxel and the mean time course of this component. Colour bar indicates z-scores (n = 24, thresholded at z > 2, uncorrected p-value < 0.0455 for a two-tailed hypothesis).

E. Reproducibility over time and between subjects

1) The individual subject z-maps obtained from dual regression of pre-stimulation

group datasets were thresholded to z > 2 using the command:

fslmaths <z-transformed file> -thr 2 <thresholded dataset>

2) The thresholded dataset was then binarised by running the command below in a

terminal. All data above 0 were given a value of 1.

fslmaths <thresholded dataset> -bin <binarised dataset>

3) To visualise reproducibility over time, the binarised datasets for each timepoint

were added to give a combined dataset for each animal. The following command

line was used:

fslmaths <binarised dataset 1> -add <binarised dataset 2> <1 plus 2>


APPENDIX 3: BRAIN ATLAS

A


B


Orbital cortex DLO, LO, MO, VO

Cingulate cortex Cg2

Auditory cortex Au1, AuD, AuV

Somatosensory cortex S1BF, S1FL, S1DZ, S1J, S1Tr, S1ULp, S2

Striatum/caudate putamen CPu

Retrosplenial cortex RSA, RSGa, RSGb

Temporal association cortex TeA

Prelimbic cortex PrL

Parasubiculum PaS

Entorhinal cortex Dsc, LEnt, Ment

Hippocampus alv, CA1, CA2, CA3, fi

Visual cortex V1B, V1M, V2L, V2ML, V2MM

Inferior colliculus BIC, bic, CIC, cic, DCIC, ECIC, ReIC

Motor cortex M1, M2

Supplementary Figure 4. Snapshots of the rat brain atlas corresponding to coordinates y=71 and z=46 (A) and y=60 and z=40 (B) in the pre-stimulation group-ICA components 0 and 11 respectively. The brain regions considered and their respective abbreviations are given. (Paxinos

and Watson 1998)

Resting state fMRI study of brain activation using rTMS in ... · We used independent component...

Documents

Transcript of Resting state fMRI study of brain activation using rTMS in ... · We used independent component...