Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau,...
Transcript of Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau,...
![Page 1: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/1.jpg)
Bayesian Inference in fMRI
Will Penny Bayesian Approaches in Neuroscience
Karolinska Institutet, Stockholm
February 2016
![Page 2: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/2.jpg)
![Page 3: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/3.jpg)
![Page 4: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/4.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 5: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/5.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 6: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/6.jpg)
Bayes Rule for Gaussians
Likelihood and Prior
(1) (1) (1)
(1) (2) (2)
( | ) ( , )
( ) ( , )
p y N
p N
(1)
(1) (2)
(1) (2)(1) (2)
( | ) ( , )p y N m P
P
mP P
Posterior
)2( m )1(
Relative Precision Weighting
Prior
Likelihood Posterior
![Page 7: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/7.jpg)
Global Shrinkage Priors
observations
GLM
prior precision
of GLM coeff
Observation
noise
Y
INp 1,0
XY
K.J. Friston and W.D. Penny. Posterior probability maps and SPMs.
NeuroImage, 19(3):1240-1249, 2003.
![Page 8: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/8.jpg)
ths
)|( yp thth pysp )|(
Posterior distribution: probability of the effect given the data
Posterior Probability Map: images of the probability that an
activation exceeds some specified threshold sth, given the
data y
)|( yp
Two thresholds:
• activation threshold sth : percentage of whole brain mean
signal (physiologically relevant size of effect)
• probability pth that voxels must exceed to be displayed
(e.g. 95%)
Posterior
![Page 9: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/9.jpg)
Mean (Cbeta_*.img)
Std dev (SDbeta_*.img)
activation
threshold
ths
Posterior density q(βn)
Probability mass pn
probability of getting an effect, given the data
),()( nnn Nq mean: size of effect
covariance: uncertainty
thpp
Display only voxels
that exceed e.g. 95%
PPM (spmP_*.img)
)( thsqp
PPM
![Page 10: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/10.jpg)
Choice of Priors
L M Harrison, W Penny, J Daunizeau, and K J Friston.
Diffusion-based spatial priors for functional magnetic resonance
images. Neuroimage, 41(2):408-23, 2008.
Nonstationary smoothness:
W.D. Penny, N. Trujillo-Barreto, and K.J. Friston. Bayesian fMRI time
series analysis with spatial priors. NeuroImage, 24(2):350-362, 2005.
Stationary smoothness:
K.J. Friston and W.D. Penny. Posterior probability maps and SPMs.
NeuroImage, 19(3):1240-1249, 2003.
Global Shrinkage:
![Page 11: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/11.jpg)
AR coeff
(correlated noise)
prior precision
of AR coeff
A
Posterior ML
aMRI Smooth Y
Stationary Smoothness Priors
observations
GLM
prior precision
of GLM coeff
Observation
noise
Y
LNp 1,0 XY
![Page 12: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/12.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 13: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/13.jpg)
Canonical
K Friston et al. Event-Related fMRI: Characterizing differential
responses, Neuroimage 7, 30-40, 1998
Two Gamma functions
fitted to data from
auditory cortex.
“Canonical” function
f(w,t) with w width and
t time.
![Page 14: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/14.jpg)
Canonical
Temporal
Hemodynamic Response Functions
Temporal derivative,
df/dt
![Page 15: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/15.jpg)
Canonical
Temporal
Dispersion
Hemodynamic Response Functions
Dispersion derivative,
df/dw
![Page 16: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/16.jpg)
Canonical
Temporal
Dispersion
Hemodynamic Response Functions
These three functions
together comprise an
“Informed Basis Set”
![Page 17: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/17.jpg)
Informed (Inf)
Fourier (F) Finite Impulse Response (FIR)
Gamma
![Page 18: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/18.jpg)
Hemodynamic Response Functions
R Henson et al. Face repetition effects in implicit and explicit memory tests
as measured by fMRI. Cerebral Cortex, 12:178-186.
W.D. Penny, G Flandin and N. Trujillo-Barreto. Bayesian Comparison of
Spatially Regularised General Linear Models . HBM, 28:275-293, 2007.
![Page 19: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/19.jpg)
Bayesian Model Comparison
Prior Posterior
𝑝(𝑚|𝑦) = 𝑝 𝑦 𝑚 𝑝(𝑚)
𝑝 𝑦 𝑚′ 𝑝(𝑚′)𝑚′
Log Evidence = log p(y|m)
![Page 20: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/20.jpg)
Hemodynamic Response Functions
Left Occipital Cortex: Inf-2 is the preferred model
![Page 21: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/21.jpg)
Hemodynamic Response Functions
Right Occipital Cortex: Inf-3 is the preferred model
![Page 22: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/22.jpg)
Hemodynamic Response Functions
Sensorimotor Cortex: Inf-3 is the preferred model
![Page 23: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/23.jpg)
( , , )
( )
g z
y b
x x h
x
Hemodynamic
variables
[ , , , ]s f v qx
Hemodynamic
parameters
Dynamics
K Friston. Bayesian Estimation of Dynamical Systems: An application to
fMRI, Neuroimage 16, 513-530, 2002
Seconds
![Page 24: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/24.jpg)
Hemodynamic Response Functions
R Buxton et al. Dynamics of Blood Flow and Oxygenation Changes
During brain activation: The Balloon Model , Magnetic Resonance in Medicine,
39:855-864.
![Page 25: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/25.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 26: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/26.jpg)
S. Kumar and W. Penny (2014). Estimating Neural Response Functions from
fMRI. Frontiers in Neuroinformatics, 8th May, doi: 10.3389/fninf.2014.00048.
Population Receptive Fields
K Friston et al. (2007) Variational free energy and the Laplace approximation.
Neuroimage, 34, 220–234.
![Page 27: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/27.jpg)
Gaussian Population Receptive Fields
![Page 28: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/28.jpg)
Gaussian Population Receptive Fields
![Page 29: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/29.jpg)
Mexican-Hat Population Receptive Fields
![Page 30: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/30.jpg)
Mexican-Hat Population Receptive Fields
![Page 31: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/31.jpg)
Which Parametric Function is a Better Descriptor ?
Gaussian
Mexican-Hat
![Page 32: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/32.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 33: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/33.jpg)
Computational fMRI
…
1 2 40
trials
1 4 3 2
Subjects pressed 1 of 4 buttons depending on
the category of visual stimulus.
The 4 categories of stimuli occurred with
different frequencies over a session.
Brain responses are then hypothesised to be
proportional to the surprise, S, associated
with each stimulus where S=log(1/p).
But over what time scale is the probability p
estimated ? And do different brain regions
use different time scales ?
L. Harrison, S Bestmann, M. Rosa, W. Penny and G. Green (2011). Time scales of
representation in the human brain: weighing past information to predict
future events. Frontiers in Human Neuroscience, 5, 00037.
![Page 34: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/34.jpg)
Long Time Scale (LTS)
Short Time Scale (STS)
Enter surprise as a Parametric Modulator in first level GLM analysis.
Which surprise variable (STS or LTS) underlies the best model of fMRI responses?
![Page 35: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/35.jpg)
kr
k
BMS maps
PPM
EPM
model k Compute log-evidence map
for each model/subject
model 1
model K
subject 1
subject N
)( krq
kr
941.0)5.0( krq
Probability that model k
generated data
M Rosa, S.Bestmann, L. Harrison, and W Penny. Bayesian model selection
maps for group studies. Neuroimage, Jan 1 2010; 49(1):217-24.
![Page 36: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/36.jpg)
L. Harrison, S Bestmann, M. Rosa, W. Penny and G. Green (2011). Time scales of
representation in the human brain: weighing past information to predict future events.
Frontiers in Human Neuroscience, 5, 00037.
Exceedance Probability Maps
![Page 37: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/37.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 38: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/38.jpg)
Relate Behavioural Descriptors, X, to fMRI data Y via voxel weights
As the number of voxels in a region will likely exceed the number of time
points in the fMRI time series, and only some combination of them will be useful
for prediction we need to select ‘features’
K Friston et al. (2008) Bayesian decoding of brain images.
Neuroimage, 39:181-205.
Which type of feature will be useful for decoding (1) voxels, (2) clusters,
(3) singular vectors, (4) support vectors ?
![Page 39: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/39.jpg)
Multivariate Decoding
1.Voxels
2.Clusters
Which type of feature will be useful for decoding (1) voxels, (2) clusters,
(3) singular vectors, (4) support vectors ?
![Page 40: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/40.jpg)
A Morcom and K Friston (2012) Decoding episodic memory in ageing: A Bayesian
Analysis of activity patterns predicting memory. Neuroimage 59, 1772-1782.
Clustered
Distributed
Clusters
Voxels
![Page 41: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/41.jpg)
A Morcom and K Friston (2012) Decoding episodic memory in ageing: A Bayesian
analysis of activity patterns predicting memory. Neuroimage 59, 1772-1782.
The more clustered the representation the better the memory
![Page 42: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/42.jpg)
Multivariate Decoding
Q. With what sort of neural code
is motion represented with in V5 ?
A. Voxels
Voxels Clusters
![Page 43: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/43.jpg)
Multivariate Decoding
Q. Which brain region can motion best
be decoded from: V5 or PFC ?
A. V5.
![Page 44: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/44.jpg)
Multivariate Decoding
Voxels
![Page 45: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/45.jpg)
A Maas et al (2014) Laminar activity in the
hippocampus and entorhinal cortex
related to novelty and episodic encoding
Nature Communications, 5:5547.
![Page 46: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/46.jpg)
A Maas et al (2014) Laminar activity in the
hippocampus and entorhinal cortex
related to novelty and episodic encoding
Nature Communications, 5:5547.
Novelty
Subsequent
Memory
![Page 47: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/47.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Decoding
• Dynamic Causal Modelling
Overview
![Page 48: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/48.jpg)
Single region
1 11 1 1z a z cu
u2
u1
z1
z2
z1
u1
a11 c
![Page 49: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/49.jpg)
Multiple regions
1 11 1 1
2 21 22 2 2
0
0
z a z uc
z a a z u
u2
u1
z1
z2
z1
z2
u1
a11
a22
c
a21
![Page 50: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/50.jpg)
Modulatory inputs
1 11 1 1 1
2
2 21 22 2 21 2 2
0 0 0
0 0
z a z z ucu
z a a z b z u
u2
u1
z1
z2
u2
z1
z2
u1
a11
a22
c
a21
b21
![Page 51: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/51.jpg)
Reciprocal connections
1 11 12 1 1 1
2
2 21 22 2 21 2 2
0 0
0 0
z a a z z ucu
z a a z b z u
u2
u1
z1
z2
u2
z1
z2
u1
a11
a22
c
a12
a21
b21
![Page 52: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/52.jpg)
Neurodynamics
i i
i
z Az u B z Cu
Inputs Change in
Neuronal
Activity
Neuronal
Activity
Intrinsic
Connectivity
Matrix
Modulatory
Connectivity
Matrices
Input
Connectivity
Matrix
![Page 53: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/53.jpg)
Rowe et al. 2010, Dynamic causal modelling of effective connectivity from fMRI:
Are results reproducible and sensitive to Parkinson’s disease and its treatment?
NeuroImage, 52:1015-1026.
Age-matched
controls
PD patients
on medication
PD patients
off medication
DA-dependent functional
disconnection of the SMA
Selection of action modulates
connections between PFC and SMA
![Page 54: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/54.jpg)
Brodersen et al. 2011, Generative Embedding for Model-Based
Classification of fMRI data. PLoS Comput. Biol. 7(6):e1002079.
step 2 —
kernel construction
step 1 —
model inversion
measurements
from an individual
subject
subject-specific
inverted DCM
subject representation in the
generative score space
A →
B
A →
C
B →
B
B →
C
A
C B
step 3 —
support vector classification
separating hyperplane
fitted to discriminate
between groups
A
C B
jointly discriminative
model parameters
step 4 —
interpretation
![Page 55: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/55.jpg)
Model-based decoding of disease status:
mildly aphasic patients (N=11) vs. controls (N=26)
Connectional fingerprints
from a 6-region DCM of
auditory areas during speech
perception
MGB
PT
HG
(A1)
S
MGB
PT
HG
(A1)
S
![Page 56: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/56.jpg)
Model-based decoding of disease status:
aphasic patients (N=11) vs. controls (N=26)
Classification accuracy
MGB
PT
HG
(A1)
MGB
PT
HG
(A1)
auditory stimuli
![Page 57: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/57.jpg)
-10
0
10
-0.5
0
0.5
-0.1
0
0.1
0.2
0.3
0.4
-0.4
-0.2
0 -0.5
0
0.5-0.4
-0.35
-0.3
-0.25
-0.2
-0.15
-10
0
10
-0.5
0
0.5
-0.1
0
0.1
0.2
0.3
0.4
-0.4
-0.2
0 -0.5
0
0.5-0.4
-0.35
-0.3
-0.25
-0.2
-0.15
gen
era
tive
em
bed
din
g
L.H
G
L.H
G
Vo
xel (6
4,-
24,4
) m
m
L.MGB L.MGBVoxel (-42,-26,10) mm
Voxel (-56,-20,10) mm R.HG L.HG
controlspatients
Voxel-based feature space Generative score space
Multivariate searchlight
classification analysis
Generative embedding
using DCM
![Page 58: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/58.jpg)
• Posterior Probability Maps
• Hemodynamic Response Functions
• Population Receptive Fields
• Computational fMRI
• Multivariate Bayes
• Dynamic Causal Modelling
Summary
![Page 59: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/59.jpg)
Savage-Dickey Ratios
Bayesian equivalent of inference using F-tests implemented
using Savage-Dickey approximations to the log Bayes Factor.
W. Penny and G. Ridgway (2013). Efficient Posterior Probability Mapping
using Savage-Dickey Ratios. PLoS One 8(3), e59655
![Page 60: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/60.jpg)
Faces versus scrambled faces
RFX analysis
on 18 subjects.
Data from
Rik Henson.
![Page 61: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/61.jpg)
Faces versus scrambled faces: Evidence for Null
Using command line call to spm_bms_test_null.m
Probability of
Null Hypothesis
![Page 62: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/62.jpg)
One parameter
Likelihood and Prior
(1) (1) (1)
(1) (2) (2)
( | ) ( , )
( ) ( , )
p y N
p N
(1)
(1) (2)
(1) (2)(1) (2)
( | ) ( , )p y N m P
P
mP P
Posterior
)2( m )1(
Relative Precision Weighting
Prior
Likelihood Posterior
![Page 63: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/63.jpg)
Two parameters
![Page 64: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/64.jpg)
Bayes Rule for Gaussians
Likelihood:
Prior:
Bayes rule:
![Page 65: Bayesian Inference in fMRI · 2016. 2. 4. · Choice of Priors L M Harrison, W Penny, J Daunizeau, and K J Friston. Diffusion-based spatial priors for functional magnetic resonance](https://reader035.fdocuments.net/reader035/viewer/2022063020/5fe153f825a5d664a81e7b33/html5/thumbnails/65.jpg)
Model comparison
“Occam’s razor” :
mo
de
l e
vid
en
ce
p(y
|m)
space of all data sets
y=f(
x)
y =
f(x
)
x
Model evidence: