media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements...

21
Lett et al. Prefrontal White Matter Structure Mediates the Influence of GAD1 on Working Memory Supplemental Information Supplement Methods & Materials Cortical Thickness Mapping All T1-weighted MRIs were submitted to the CIVET pipeline (v1.2). T1 images were registered to the ICBM152 nonlinear template with a 9-parameter linear transformation, intensity inhomogeneity corrected and tissue classified (for grey matter, white matter, and cerebral spinal fluid) . Deformable models were used to create white and gray matter surfaces for each hemisphere separately, resulting in 4 surfaces of 40,962 vertices each . From these surfaces, the t-link metric was derived for determining the distance between the white and gray surfaces . The thickness data were blurred using a 20-mm surface-based diffusion kernel in preparation for statistical analyses. Unnormalized, native-space thickness values were used in all analyses owing to the poor correlation between cortical thickness and brain volume . Tract-Based Spatial Statistics 1

Transcript of media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements...

Page 1: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Prefrontal White Matter Structure Mediates the Influence of GAD1 on Working Memory

Supplemental Information

Supplement Methods & Materials

Cortical Thickness Mapping

All T1-weighted MRIs were submitted to the CIVET pipeline (v1.2). T1 images were registered

to the ICBM152 nonlinear template with a 9-parameter linear transformation, intensity

inhomogeneity corrected and tissue classified (for grey matter, white matter, and cerebral spinal

fluid) . Deformable models were used to create white and gray matter surfaces for each

hemisphere separately, resulting in 4 surfaces of 40,962 vertices each . From these surfaces, the

t-link metric was derived for determining the distance between the white and gray surfaces . The

thickness data were blurred using a 20-mm surface-based diffusion kernel in preparation for

statistical analyses. Unnormalized, native-space thickness values were used in all analyses owing

to the poor correlation between cortical thickness and brain volume .

Tract-Based Spatial Statistics

All diffusion tensor imaging (DTI) analyses were done using tools implemented in the FSL

toolkit v.5.0 . The three repetitions for each subject’s 4D DW-MRI volumes were merged. FSL

eddy was applied for correcting eddy current and movements in the diffusion data. After skull

stripping using BET , fractional anisotropy (FA) images were created by fitting a tensor model at

each voxel using DTIFit. FA quantifies directionality of water diffusion on a scale from zero

(random diffusion) to one (diffusion in one direction). Voxel-wise analysis of the FA data was

carried out using Tract-Based Spatial Statistics (TBSS, v1.2) . TBSS projects all subjects'

fractional anisotropy (FA) data onto a mean FA tract skeleton, before applying voxel-wise cross-

1

Page 2: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

subject statistics. Briefly, FA images underwent nonlinear registration to the FMRIB58_FA

target image. Next, the mean FA image was iteratively generated from scans and was then

aligned to MNI 152 standard space using an affine transformation. An average white matter

skeleton was then generated from the mean of all subjects’ transformed FA images at a threshold

of 0.2. For group comparisons, each subject’s FA data was projected onto the white matter

skeleton and voxel-wise statistics were calculated using randomise (v2.1) with 10,000

permutations.

Localization of the DLPFC for TMS-EEG

Details of the DLPFC localization method have been previously described . DLPFC localization

was achieved through neuronavigation techniques using the MINIBIRD system (Ascension

Technologies) and MRIcro/registration software using a T1-weighted MRI scan obtained for

each subject with seven fiducial markers in place. Stimulation was directed at the junction of the

middle and anterior one-third of the middle frontal gyrus [Talairach coordinates (x, y, z) = −50,

30, 36] corresponding with posterior regions of Brodmann area (BA) 9 that overlaps with the

superior section of BA 46.

LICI assessment

Monophasic TMS pulses were administered using a 7-cm figure-of-8 coil, and two Magstim 200

stimulators (Magstim Company Company, Carmarthenshire, Wales) connected via a Biostim

module. TMS was administered over the DLPFC. Inhibition was measured through LICI and

indexed through electromyography and EEG at the optimal 100 ms interstimulus interval . One

hundred TMS stimuli were delivered per condition (paired and single-pulse) every 5 s. The

intensity of TMS pulses was determined at the beginning of each experiment and it was set such

2

Page 3: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

that it elicited an average motor evoked potential of 1 mV peak-to-peak upon delivery of 20

pulses over the motor cortex. Both the conditioning stimulus and test stimulus were delivered at

the same suprathreshold intensity. Previous analysis has demonstrated no significant between-

group differences were found for the 1 mV peak-to-peak TMS intensity among patients and

controls .

EEG recording and processing

Details of the EEG recording and processing have been previously described . In brief, EEG was

acquired through a 64-channel Synamps 2 EEG system. All electrodes were referenced to an

electrode positioned posterior to the Cz electrode. EEG signals were DC filtered and recorded

with a low pass filter of 100 Hz at a 20 kHz sample rate (to avoid the TMS related artifact).

The TMS-EEG signals were processed offline using MATLAB (The MathWorks Inc. Natick,

MA, USA). All signals were down sampled from 20 kHz to 1 kHz and segmented with respect to

the test stimulus such that each epoch included 1000 ms pre-stimulus baseline and 2000 ms post-

stimulus activity. Epochs were baseline corrected with respect to the TMS-free pre-stimulus

interval. The segment from -100 ms to 10 ms was removed from the TMS pulse removing all

single and paired pulse stimulation. EEGs were visually inspected to eliminate trials and

channels highly contaminated by artefacts including: muscle activity, 60 Hz noise, and

movement-related activity. Two independent component analyses (ICA) were applied. The initial

ICA was employed to remove the typical TMS-related decay artefact. Subsequently, a bandpass

FIR filter was applied (1-55 Hz). The second ICA was used to remove eye-related artefacts and

remaining muscle components.

Voxel-wide mediation analysis of the TBSS skeleton

3

Page 4: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Voxel-wise mediation analysis was performed in MATLAB (R2013b). We used the multiple

regression approach described by Baron and Kenny , and applied this approach across the entire

TBSS FA skeleton. A 4D image TBSS skeleton of the subjects was imported into MATLAB,

and transformed into an array of all non-zero FA voxels across each subject (121023 voxels were

extracted). We regressed out the effects of diagnosis, sex, and age for all voxels. Our mediation

analysis was accomplished with three regression equations applied across all voxels. First, we

regressed the independent variable (GAD1 genotype) against white matter FA. A z-score was

produced for each non-zero voxel and was used to produce a 3D image of z-scores (‘Path A’).

We then applied TFCE in the FSL ‘fslmaths’ function with E=2, H=1, and the neighbourhood-

connectivity parameter = 26 as recommended in TBSS analysis . 10,000 permutations (i.e.

randomization analysis) were then performed and the maximum z-statistics for each permutation

was used to assess significance accounting for FWE. Second, we regressed the mediator variable

(white matter FA at each voxel) against cognitive performance at each voxel (‘Path B’). A 3D

image of z-scores was produced, and we tested significance using the same TFCE and

permutation test technique. Third, we regressed the independent variable (GAD1 genotype) on

cognitive performance (‘Path C’). A significant association in all three sets of regressions then

allowed us to proceed with the Sobel equation to assess the indirect effect of the independent

variable on the dependent variable via the mediator (at each voxel of white matter FA skeleton).

We used the unstandardized regression coefficients (beta) and the standard errors (SE) from

‘Path A’ and ‘Path B’ in order to produce a z-value at each white matter FA voxel (Sobel

equation: z-value = beta(Path A)*beta(Path B)/ √(beta(path B)2 *SE(Path A)

2 + beta(Path A)2 *SE(Path B)

2)). A 3D

image of z-values were produced, and we applied TFCE. Significant mediation was assessed

using the max TFCE transformed z-value from each 10,000 permutations. TFCE transformed

4

Page 5: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Sobel z-value voxels that were greater than 95% of the max TFCE transformed z-values were

deemed significant (i.e., pFWE-corrected<0.05).Voxels that were It should be noted that randomization

strategies to assess significance of the Sobel equation are considered to be a better alternative

than parameter tests that impose distribution assumptions .

Assessment of Working Memory and Stroop interference

All subjects underwent a battery of cognitive tests administered over approximately 1.5 hours.

This battery includes a wide range of cognitive domains with varying degrees of impairment in

schizophrenia , and has been previously described . We chose two working memory span tasks

(verbal working memory: the letter-number sequencing task (LNS); non-verbal working

memory: the digit-span forward task (digit-span)) . The LNS requires an understanding of order

of the stimuli related to previous learning, whereas digit-span requires on the repetition of the

forward order. We further assessed selective attention using the Stroop Neuropsychological

Screening Test . We assessed the Stroop interference effect by using the reaction time of the

colour-word task (time per item), and a ratio score (i.e. the Stroop difference score divided by the

latency to colour-word control items). This ratio score provides a more conservative estimate of

Stroop interference because it controls for differences in overall response latencies, both between

and within groups .

Supplemental Results

GAD1 and Cortical Thickness Regions of Interests.

Considering the previous significant association with parahippocampal cortical thickness , we

performed an exploratory analysis with mean thickness values of 52 regions parcellated from the

5

Page 6: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Probabilistic Brain Atlas . After Bonferroni correction, T allele risk carriers had significantly

lower cortical thickness, particularly in the right temporal lobe area (Supplementary Table S3).

Figure S1. Association of GAD1 rs3749034 genotype with mean TBSS FA of the significant voxels from the largest cluster (7688 voxels; max voxel [104x162x79]; pFWE=0.03) in patients and healthy controls. GAD1 rs3749034 genotype is associated with mean FA (F1,195=9.65, p=1.02x10-4). Hollow circles represent each individual. Mean and 95% confidence intervals for each group are to the immediate right. Mean fractional anisotropy (FA) is corrected for sex and age (46.3 years). CNT, healthy control; DLPFC, dorsolateral prefrontal cortex. FA, fractional anisotropy, TBSS, tract-based spatial statistics; SCZ, patients with schizophrenia;

6

Page 7: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Figure S2. The association between GAD1 genotype and working memory related performance in healthy controls and schizophrenia patients. Hollow circles represent each subject with C-allele homozygotes and solid circles represent each T-allele carriers. Mean and 95% confidence intervals for each group are to the immediate right. Cognitive performance for each task has been corrected for age (46.3 years) and IQ (WTAR: 115.1). (a) There is a significant effect of GAD1 genotype on Digit-span performance (F1,188=8.0, p=0.005). (b) Genotype predicted a nominal effect LNS performance (F1,188=5.0,p=0.03). (c) Genotype was not associated with Stroop (Time/Item). (d) Genotype was associated with Stroop (Ratio) score (F1,188=7.0, p=0.009). Diagnosis was associated with performance in all tasks. No genotype-by-diagnosis interactions were observed. CNT, healthy control; Digit-Span, digit-span forward performance; GAD1, glutamate decarboxylase 1 (brain, 67kDa); LNS, letter-number sequencing performance; SCZ, Schizophrenia patients.

7

Page 8: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Figure S3. Higher TBSS skeleton white matter FA correlates with better digit span performance. Areas in yellow correspond to p<0.05 after correction for family-wise error. Green represents the mean FA skeleton overlaid on the FMRIB58_FA standard space image.

8

Page 9: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Tables

Table S1. Demographics of TMS-EEG SampleControls (N= 115) Schizophrenia (N = 80)Mean SD Mean SD T54 P

Age (years) 34.58 10.93 34.59 10.39 0.12 0.99Count Frequency Count Frequency χ2 p

Handedness (R) 33 100% 23 100%Sex (M) 15 45.5% 18 78.3% 6.02 0.02

Table S2. Count and frequency of GAD1 rs3749034 genotypes by diagnosis in the imaging-genetics sample and the TMS-EEG sample.

Imaging-Genetics (N=195) TMS-EEG (N=56)Genotype Count Frequency (%) Count Frequency (%)

Healthy Controls CC 69 60.00 20 60.61CT 41 35.65 11 33.33TT 5 4.35 2 6.06

Schizophrenia CC 39 48.75 10 43.48CT 34 42.50 11 47.83TT 7 8.75 2 8.70

Table S3. Exploratory analysis for association among LONI probabilistic brain atlas cortical thickness regions and GAD1 rs3749034 genotype, diagnosis, and their interaction.

GAD1 Genotype Diagnosis Genotype*DiagnosisCortical Thickness Region F1,195 PFDR-corrected F1,195 PFDR-corrected F1,195 PFDR-corrected

rostral middle frontal gyrus inferior left 5.9 7.15E-02 25.4 2.60E-05 0.4 9.82E-01caudal middle frontal gyrus left 7.4 5.18E-02 12.3 1.26E-03 0.3 9.82E-01superior frontal gyrus left 3.9 1.35E-01 11.2 1.76E-03 0.0 9.82E-01rostral middle frontal gyrus superior left 4.4 1.22E-01 23.9 3.47E-05 0.0 9.82E-01inferior frontal gyrus left 3.8 1.35E-01 14.1 8.08E-04 0.0 9.82E-01precentral gyrus left 1.4 3.20E-01 4.7 3.73E-02 2.3 8.38E-01middle orbitofrontal gyrus left 2.5 2.18E-01 13.7 9.30E-04 0.1 9.82E-01lateral orbitofrontal gyrus left 0.3 6.41E-01 6.7 1.29E-02 0.0 9.82E-01gyrus rectus left 3.0 1.80E-01 10.1 2.64E-03 0.0 9.82E-01postcentral gyrus left 3.8 1.35E-01 1.7 2.00E-01 0.6 9.82E-01superior parietal gyrus left 2.5 2.18E-01 3.3 7.81E-02 2.7 8.38E-01supramarginal gyrus left 4.2 1.26E-01 8.7 4.82E-03 0.2 9.82E-01angular gyrus left 2.0 2.54E-01 8.8 4.56E-03 0.0 9.82E-01precuneus left 3.4 1.55E-01 10.0 2.69E-03 0.1 9.82E-01superior occipital gyrus left 0.2 6.76E-01 2.8 1.00E-01 3.5 8.38E-01middle occipital gyrus left 0.3 6.21E-01 10.2 2.62E-03 0.1 9.82E-01

9

Page 10: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

inferior occipital gyrus left 2.9 1.82E-01 11.6 1.59E-03 0.2 9.82E-01cuneus left 0.4 5.99E-01 4.6 3.75E-02 2.4 8.38E-01superior temporal gyrus left 3.6 1.43E-01 15.2 6.29E-04 1.7 9.44E-01middle temporal gyrus left 6.1 6.81E-02 20.8 1.17E-04 0.0 9.82E-01inferior temporal gyrus left 8.2 4.76E-02 12.2 1.30E-03 0.1 9.82E-01parahippocampal gyrus left 0.7 5.09E-01 11.1 1.82E-03 0.5 9.82E-01lingual gyrus left 1.9 2.58E-01 26.0 2.60E-05 1.4 9.82E-01fusiform gyrus left 2.3 2.41E-01 14.2 8.08E-04 1.0 9.82E-01insular cortex left 0.2 6.52E-01 19.0 1.91E-04 0.3 9.82E-01cingulate gyrus left 0.0 8.78E-01 1.7 1.92E-01 0.0 9.82E-01rostral middle frontal gyrus inferior right 6.5 5.90E-02 14.4 8.08E-04 0.2 9.82E-01caudal middle frontal gyrus right 6.6 5.90E-02 10.7 2.06E-03 1.2 9.82E-01superior frontal gyrus right 4.5 1.22E-01 13.3 9.33E-04 0.0 9.82E-01rostral middle frontal gyrus superior right 1.9 2.58E-01 10.0 2.69E-03 0.0 9.82E-01inferior frontal gyrus right 5.0 9.63E-02 12.7 1.12E-03 0.2 9.82E-01precentral gyrus right 2.2 2.47E-01 2.8 1.00E-01 0.7 9.82E-01middle orbitofrontal gyrus right 1.2 3.61E-01 13.2 9.52E-04 0.0 9.82E-01lateral orbitofrontal gyrus right 0.7 5.09E-01 7.5 8.91E-03 0.1 9.82E-01gyrus rectus right 7.2 5.28E-02 17.3 3.51E-04 2.2 8.38E-01postcentral gyrus right 2.0 2.51E-01 5.0 3.16E-02 0.5 9.82E-01superior parietal gyrus right 1.5 3.11E-01 3.5 7.19E-02 2.7 8.38E-01supramarginal gyrus right 7.4 5.18E-02 14.1 8.08E-04 0.1 9.82E-01angular gyrus right 2.1 2.51E-01 13.4 9.33E-04 0.0 9.82E-01precuneus right 1.5 3.15E-01 10.9 1.94E-03 0.1 9.82E-01superior occipital gyrus right 0.7 5.09E-01 2.1 1.50E-01 3.3 8.38E-01middle occipital gyrus right 0.4 5.94E-01 11.2 1.76E-03 0.8 9.82E-01inferior occipital gyrus right 5.3 9.21E-02 12.5 1.20E-03 2.1 8.38E-01cuneus right 0.5 5.60E-01 9.6 3.17E-03 3.8 8.38E-01superior temporal gyrus right 8.6 4.76E-02 19.5 1.77E-04 0.1 9.82E-01middle temporal gyrus right 9.9 3.29E-02 17.1 3.51E-04 0.6 9.82E-01inferior temporal gyrus right 10.7 3.26E-02 11.5 1.60E-03 0.9 9.82E-01parahippocampal gyrus right 2.9 1.82E-01 15.2 6.29E-04 1.7 9.44E-01lingual gyrus right 3.7 1.40E-01 16.2 4.74E-04 0.8 9.82E-01fusiform gyrus right 11.1 3.26E-02 13.4 9.33E-04 0.1 9.82E-01insular cortex right 0.4 5.94E-01 11.8 1.51E-03 0.0 9.82E-01cingulate gyrus right 0.6 5.12E-01 6.2 1.73E-02 0.1 9.82E-01

* Age and sex are included as covariates. P-values are corrected for false discovery rate (FDR) using the Benjamini and Hochberg method.

Table S4. Partial correlation matrix among Digit-Span, LNS, Stroop Time/Item, and Stroop Ratio performance scores in heathy controls and schizophrenia patients.Healthy Controls Digit- LNS Stroop Stroop

10

Page 11: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Span (Time/Item) (Ratio)Digit-Span r .542 .016 .068

p   .000 .861 .464df 116 116 116

LNS r .542 -.081 .031p .000   .384 .738df 116 116 116

Stroop (Time/Item)

r .016 -.081 .387p .861 .384   .000df 116 116 116

Stroop (Ratio) r .068 .031 .387p .464 .738 .000  df 116 116 116

Schizophrenia Patients Digit-Span

LNS Stroop (Time/Item)

Stroop (Ratio)

Digit-Span r .368 -.013 .179p   .001 .912 .132df 70 70 70

LNS r .368 -.291 -.161p .001   .013 .177df 70 70 70

Stroop (Time/Item)

r -.013 -.291 .849p .912 .013   .000df 70 70 70

Stroop (Ratio) r .179 -.161 .849p .132 .177 .000  df 70 70 70

* Performance scores are corrected for age (46.3 years), and WTAR (115.1). Digit-Span, digit-span forward performance; df, degrees of freedom; LNS, letter-number sequencing performance; p, p-value; r, correlation coefficient.

References

11

Page 12: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Ad-Dab'bagh Y, Singh V, Robbins S, Lerch J, Lyttleton O, Fombonne E, et al (eds) (2005). Native space cortical thickness measurement and the absence of correlation to cerebral volume. Organization of Human Brain Mapping. Toronto. Neuroimage: Toronto.

Baron RM, Kenny DA (1986). The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 51(6): 1173-1182.

Brauns S, Gollub RL, Walton E, Hass J, Smolka MN, White T, et al (2013). Genetic variation in GAD1 is associated with cortical thickness in the parahippocampal gyrus. Journal of psychiatric research 47(7): 872-879.

Collins DL, Neelin P, Peters TM, Evans AC (1994). Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. Journal of computer assisted tomography 18(2): 192-205.

Garcia Dominguez L, Radhu N, Farzan F, Daskalakis ZJ (2014). Characterizing long interval cortical inhibition over the time-frequency domain. PloS one 9(3): e92354.

Golden CJ, Freshwater SM (1978). Stroop color and word test.

Hale JB, Hoeppner J-AB, Fiorello CA (2002). Analyzing digit span components for assessment of attention processes. Journal of Psychoeducational Assessment 20(2): 128-143.

Kim JS, Singh V, Lee JK, Lerch J, Ad-Dab'bagh Y, MacDonald D, et al (2005). Automated 3-D extraction and evaluation of the inner and outer cortical surfaces using a Laplacian map and partial volume effect classification. Neuroimage 27(1): 210-221.

Lerch JP, Evans AC (2005). Cortical thickness analysis examined through power analysis and a population simulation. Neuroimage 24(1): 163-173.

Perlstein WM, Carter CS, Barch DM, Baird JW (1998). The Stroop task and attention deficits in schizophrenia: a critical evaluation of card and single-trial Stroop methodologies. Neuropsychology 12(3): 414-425.

Plude DJ, Hoyer WJ (1981). Adult age differences in visual search as a function of stimulus mapping and processing load. Journal of gerontology 36(5): 598-604.

Preacher KJ, Hayes AF (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior research methods 40(3): 879-891.

Radhu N, Garcia Dominguez L, Farzan F, Richter MA, Semeralul MO, Chen R, et al (2015). Evidence for inhibitory deficits in the prefrontal cortex in schizophrenia. Brain 138(Pt 2): 483-497.

12

Page 13: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Rajji TK, Ismail Z, Mulsant BH (2009). Age at onset and cognition in schizophrenia: meta-analysis. Br J Psychiatry 195(4): 286-293.

Randolph C, Tierney MC, Mohr E, Chase TN (1998). The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): preliminary clinical validity. Journal of clinical and experimental neuropsychology 20(3): 310-319.

Rusjan PM, Barr MS, Farzan F, Arenovich T, Maller JJ, Fitzgerald PB, et al (2010). Optimal transcranial magnetic stimulation coil placement for targeting the dorsolateral prefrontal cortex using novel magnetic resonance image-guided neuronavigation. Human brain mapping 31(11): 1643-1652.

Sanger TD, Garg RR, Chen R (2001). Interactions between two different inhibitory systems in the human motor cortex. The Journal of physiology 530(Pt 2): 307-317.

Shattuck DW, Mirza M, Adisetiyo V, Hojatkashani C, Salamon G, Narr KL, et al (2008). Construction of a 3D probabilistic atlas of human cortical structures. Neuroimage 39(3): 1064-1080.

Sled JG, Zijdenbos AP, Evans AC (1998). A nonparametric method for automatic correction of intensity nonuniformity in MRI data. IEEE Trans Med Imaging 17(1): 87-97.

Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE, Mackay CE, et al (2006). Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage 31(4): 1487-1505.

Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TEJ, Johansen-Berg H, et al (2004). Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 23: S208-S219.

Smith SM, Nichols TE (2009). Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 44(1): 83-98.

Smith SM, Zhang Y, Jenkinson M, Chen J, Matthews PM, Federico A, et al (2002). Accurate, robust, and automated longitudinal and cross-sectional brain change analysis. Neuroimage 17(1): 479-489.

Sobel ME (1982). Asymptotic confidence intervals for indirect effects in structural equation models. Sociological methodology 13(1982): 290-312.

Sobel ME (1986). Some new results on indirect effects and their standard errors in covariance structure models. Sociological methodology: 159-186.

Stroop JR (1935). Studies of interference in serial verbal reactions. Journal of experimental psychology 18(6): 643.

13

Page 14: media.nature.com€¦  · Web viewFSL eddy was applied for correcting eddy current and movements in the diffusion data. After skull stripping using BET (Smith et al, 2002), fractional

Lett et al.

Tohka J, Zijdenbos A, Evans A (2004). Fast and robust parameter estimation for statistical partial volume models in brain MRI. Neuroimage 23(1): 84-97.

Voineskos AN, Felsky D, Kovacevic N, Tiwari AK, Zai C, Chakravarty MM, et al (2013). Oligodendrocyte genes, white matter tract integrity, and cognition in schizophrenia. Cerebral cortex 23(9): 2044-2057.

Voineskos AN, Rajji TK, Lobaugh NJ, Miranda D, Shenton ME, Kennedy JL, et al (2012). Age-related decline in white matter tract integrity and cognitive performance: a DTI tractography and structural equation modeling study. Neurobiology of aging 33(1): 21-34.

Zijdenbos AP, Forghani R, Evans AC (2002). Automatic "pipeline" analysis of 3-D MRI data for clinical trials: application to multiple sclerosis. IEEE Trans Med Imaging 21(10): 1280-1291.

14