dynamic causal modelling for fmri - fil | ucl€¦ · dynamic causal modelling for fmri friday 18th...
TRANSCRIPT
![Page 1: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/1.jpg)
Dynamic causal modelling
for fMRI
Friday 18th May 2012
Statistical Parametric Mapping
for fMRI
2012 course
André Marreiros
![Page 2: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/2.jpg)
Overview
Brain connectivity: types & definitions
Anatomical connectivity Functional connectivity Effective connectivity
Dynamic causal models (DCMs)
Neuronal model Hemodynamic model
Estimation: Bayesian framework
Applications & extensions of DCM to fMRI data
![Page 3: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/3.jpg)
Functional specialization Functional integration
Principles of Organisation
![Page 4: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/4.jpg)
Structural, functional & effective connectivity
• anatomical/structural connectivity
= presence of axonal connections
• functional connectivity
= statistical dependencies between regional time series
• effective connectivity
= causal (directed) influences between neurons or neuronal populations
Sporns 2007, Scholarpedia
MECHANISM-FREE
MECHANISTIC MODEL
![Page 5: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/5.jpg)
Anatomical connectivity
Definition:
presence of axonal connections
• neuronal communication via synaptic contacts
• Measured with
– tracing techniques
– diffusion tensor imaging (DTI)
![Page 6: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/6.jpg)
Knowing anatomical connectivity is not enough...
• Context-dependent recruiting of connections :
– Local functions depend on network activity
• Connections show synaptic plasticity
– change in the structure and transmission properties of a synapse
– even at short timescales
Look at functional and effective connectivity
![Page 7: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/7.jpg)
Definition: statistical dependencies between regional time series
• Seed voxel correlation analysis
• Coherence analysis
• Eigen-decomposition (PCA, SVD)
• Independent component analysis (ICA)
• any technique describing statistical dependencies amongst
regional time series
Functional connectivity
![Page 8: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/8.jpg)
Seed-voxel correlation analyses
• hypothesis-driven choice of a
seed voxel
• extract reference
time series
• voxel-wise correlation with
time series from all other
voxels in the brain
seed voxel
![Page 9: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/9.jpg)
Pros & Cons of functional connectivity analysis
• Pros: – useful when we have no experimental control over
the system of interest and no model of what caused the data (e.g. sleep, hallucinations, etc.)
• Cons: – interpretation of resulting patterns is difficult / arbitrary
– no mechanistic insight
– usually suboptimal for situations where we have a priori knowledge / experimental control
Effective connectivity
![Page 10: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/10.jpg)
Effective connectivity
Definition: causal (directed) influences between neurons or neuronal populations
• In vivo and in vitro stimulation and recording
•
• Models of causal interactions among neuronal populations
– explain regional effects in terms of interregional connectivity
![Page 11: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/11.jpg)
Some models for computing effective connectivity
from fMRI data
• Structural Equation Modelling (SEM) McIntosh et al. 1991, 1994; Büchel & Friston 1997; Bullmore et al. 2000
• regression models
(e.g. psycho-physiological interactions, PPIs) Friston et al. 1997
• Volterra kernels Friston & Büchel 2000
• Time series models (e.g. MAR, Granger causality) Harrison et al. 2003, Goebel et al. 2003
• Dynamic Causal Modelling (DCM) bilinear: Friston et al. 2003; nonlinear: Stephan et al. 2008
![Page 12: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/12.jpg)
Psycho-physiological interaction (PPI)
• bilinear model of how the psychological context A changes
the influence of area B on area C :
B x A C
• A PPI corresponds to differences in regression slopes for different contexts.
![Page 13: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/13.jpg)
Psycho-physiological interaction (PPI)
We can replace one main effect in
the GLM by the time series of an
area that shows this main effect.
Task factor
Task A Task B S
tim
1
Stim
2
Stim
ulu
s fa
cto
r
A1 B1
A2 B2
e
βVTT
βV
TT y
BA
BA
3
2
1
1 )(
1
)(
e
βSSTT
βSS
TT y
BA
BA
321
221
1
)( )(
)(
)(
GLM of a 2x2 factorial design:
main effect of task
main effect of stim. type
interaction
main effect of task
V1 time series main effect of stim. type
psycho- physiological interaction
Friston et al. 1997, NeuroImage
![Page 14: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/14.jpg)
V1
attention
no attention
V1 activity
V5 a
ctivity
SPM{Z}
time
V5 a
ctivity
Friston et al. 1997, NeuroImage
Büchel & Friston 1997, Cereb. Cortex
V1 x Att.
=
V5
V5
Attention
Example PPI: Attentional modulation of
V1→V5CC
![Page 15: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/15.jpg)
PPI: Interpretation
Two possible
interpretations
of the PPI term:
V1
Modulation of V1V5
by attention
Modulation of the impact
of attention on V5 by V1
V1 V5 V1 V5
attention
V1
attention
e
βVTT
βV
TT y
BA
BA
3
2
1
1 )(
1
)(
![Page 16: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/16.jpg)
Pros & Cons of PPIs
• Pros:
– given a single source region, we can test for its context-dependent
connectivity across the entire brain
– easy to implement
• Cons:
– only allows to model contributions from a single area
– operates at the level of BOLD time series (SPM 99/2).
SPM 5/8 deconvolves the BOLD signal to form the proper interaction term,
and then reconvolves it.
– ignores time-series properties of the data
Dynamic Causal Models
needed for more robust statements of effective connectivity.
![Page 17: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/17.jpg)
Dynamic causal models (DCMs)
Basic idea Neuronal model Hemodynamic model Parameter estimation, priors & inference
Brain connectivity: types & definitions
Anatomical connectivity Functional connectivity Effective connectivity
Overview
Applications & extensions of DCM to fMRI data
![Page 18: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/18.jpg)
Basics of Dynamic Causal Modelling
DCM allows us to look at how areas within a network interact:
Investigate functional integration & modulation of specific cortical pathways
– Temporal dependency of activity within and between areas (causality)
![Page 19: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/19.jpg)
Temporal dependence and causal relations
Seed voxel approach, PPI etc. Dynamic Causal Models
timeseries (neuronal activity)
![Page 20: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/20.jpg)
Basics of Dynamic Causal Modelling
DCM allows us to look at how areas within a network interact:
Investigate functional integration & modulation of specific cortical pathways
– Temporal dependency of activity within and between areas (causality)
– Separate neuronal activity from observed BOLD responses
![Page 21: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/21.jpg)
• Cognitive system is modelled at its underlying
neuronal level (not directly accessible for fMRI).
• The modelled neuronal dynamics (Z) are
transformed into area-specific BOLD signals (y) by
a hemodynamic model (λ).
λ
Z
y
The aim of DCM is to estimate parameters
at the neuronal level such that the modelled
and measured BOLD signals are optimally
similar.
Basics of DCM:
Neuronal and BOLD level
![Page 22: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/22.jpg)
Neuronal systems are represented by
differential equations
Input u(t)
connectivity parameters
system
z(t) state
State changes of the system
states are dependent on:
– the current state z
– external inputs u
– its connectivity θ
– time constants & delays
A System is a set of elements
zn(t) which interact in a spatially
and temporally specific fashion
),,( uzFdt
dz
![Page 23: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/23.jpg)
DCM parameters = rate constants
11
dzsz
dt
Half-life
Generic solution to the ODEs in DCM:
ln 2 /s
1 1
1
( ) 0.5 (0)
(0)exp( )
z z
z s
1 1 1( ) (0)exp( ), (0) 1z t z st z z1
s
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0
0.2
0.4
0.6
0.8
1
s/2ln
)0(5.0 1z
Decay function
![Page 24: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/24.jpg)
DCM parameters = rate constants
11
dzsz
dt
Generic solution to the ODEs in DCM:
1 1 1( ) (0)exp( ), (0) 1z t z st z z1
s
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0
0.2
0.4
0.6
0.8
1
s/2ln
)0(5.0 1z
Decay function
If AB is 0.10 s-1 this
means that, per unit time,
the increase in activity in
B corresponds to 10% of
the activity in A
A
B
0.10
![Page 25: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/25.jpg)
Linear dynamics: 2 nodes
1 1
2 21 1 2
1
2
1
2 21
21
(0) 1
(0) 0
( ) exp( )
( ) exp( )
0
z sz
z s a z z
z
z
z t st
z t sa t st
a
1;4 21 as
2;4 21 as
1;8 21 as
z2
21a
z1
s
s
z1
sa21t z2
![Page 26: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/26.jpg)
Neurodynamics: 2 nodes with input
u2
u1
z1
z2
activity in z2 is coupled to z1 via coefficient a21
u1
21a
001
01211
2
1
212
1
au
c
z
z
as
z
z
z1
z2
![Page 27: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/27.jpg)
Neurodynamics: positive modulation
000
00
1
012
211
2
1
2
21
2
2
1
212
1
bu
c
z
z
bu
z
z
as
z
z
u2
u1
z1
z2
modulatory input u2 activity through the coupling a21
u1
u2
index, not squared
z1
z2
![Page 28: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/28.jpg)
Neurodynamics: reciprocal connections
0,,00
00
1
12
2121121
2
1
2
21
2
2
1
21
12
2
1
baau
c
z
z
bu
z
z
a
as
z
z
u2
u1
z1
z2
reciprocal connection
disclosed by u2
u1
u2 z1
z2
![Page 29: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/29.jpg)
0 20 40 60
0
2
4
0 20 40 60
0
2
4
seconds
Haemodynamics: reciprocal connections
blue: neuronal activity
red: bold response
h1
h2
u1
u2 z1
z2
h(u,θ) represents the BOLD response (balloon model) to input
BOLD
(without noise)
BOLD
(without noise)
![Page 30: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/30.jpg)
0 20 40 60
0
2
4
0 20 40 60
0
2
4
seconds
Haemodynamics: reciprocal connections
BOLD
with
Noise added
BOLD
with
Noise added
y1
y2
blue: neuronal activity
red: bold response
u1
u2 z1
z2
euhy ),( y represents simulated observation of BOLD response, i.e. includes noise
![Page 31: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/31.jpg)
Bilinear state equation in DCM for fMRI
state
changes connectivity
external
inputs
state
vector
direct
inputs
CuzBuAzm
j
j
j
)(1
mnmn
m
n
m
j j
nn
j
n
j
n
j
j
nnn
n
n u
u
cc
cc
z
z
bb
bb
u
aa
aa
z
z
1
1
1111
1
1
111
1
1111
modulation of
connectivity
n regions m drv inputs m mod inputs
![Page 32: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/32.jpg)
BOLD y
y
y
haemodynamic
model
Input
u(t)
activity
z2(t)
activity
z1(t)
activity
z3(t)
effective connectivity
direct inputs
modulation of
connectivity
The bilinear model CuzBuAz j
j )(
c1 b23
a12
neuronal
states
λ
z
y
integration
Neuronal state equation ),,( nuzFz Conceptual
overview
Friston et al. 2003,
NeuroImage
u
z
u
FC
z
z
uuz
FB
z
z
z
FA
jj
j
2
![Page 33: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/33.jpg)
The hemodynamic “Balloon” model
, , , ,h
sf
tionflow induc
q/vvf,Efqτ /α
dHbchanges in
1)( /αvfvτ
volumechanges in
1
)1( fγszs
ry signalvasodilato
( )
neuronal input
z t
,)(
signal BOLD
qvty
5 haemodynamic parameters
Friston et al. 2000,
NeuroImage
Stephan et al. 2007,
NeuroImage
Region-specific HRFs!
![Page 34: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/34.jpg)
fMRI data
Posterior densities
of parameters
Neuronal
dynamics Hemodynamics
Model
selection
DCM roadmap
Model inversion
using
Expectation-maximization
State space
Model
Priors
![Page 35: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/35.jpg)
Constraints on •Haemodynamic parameters
•Connections
Models of •Haemodynamics in a single region
•Neuronal interactions
Bayesian estimation
)(p
)()|()|( pypyp
)|( yp
posterior
prior likelihood term
Estimation: Bayesian framework
![Page 36: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/36.jpg)
sf
(rCBF)induction -flow
s
v
f
stimulus function u
modeled
BOLD response
v
q q/vvf,Efqτ /α1)(
dHbin changes
/αvfvτ 1
in volume changes
f
q
)1(
signalry vasodilatodependent -activity
fγszs
s
( , , )h x u ( , , )y h x u X e
observation model
hidden states },,,,{ qvfszx
state equation ( , , )x F x u
parameters
},{
},...,{
},,,,{
1
nh
mn
h
CBBA
Overview:
parameter estimation
ηθ|y
neuronal state
equation CuzBuAz j
j )(
• Specify model (neuronal and haemodynamic level)
• Make it an observation model by adding measurement error e and confounds X (e.g. drift).
• Bayesian parameter estimation using expectation-maximization.
• Result: (Normal) posterior parameter distributions, given by mean ηθ|y and Covariance Cθ|y.
![Page 37: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/37.jpg)
0 20 40 60
-10123
0 20 40 60-10123
seconds
21a
Input coupling, c1
1c
Prior density Posterior density true values
Parameter estimation: an example
u1
21a
z1
z2
Simulated response
![Page 38: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/38.jpg)
Bayesian single subject analysis
• The model parameters are
distributions that have a mean ηθ|y
and covariance Cθ|y.
– Use of the cumulative normal
distribution to test the probability
that a certain parameter is above a
chosen threshold γ:
ηθ|y
Classical frequentist test across groups
• Test summary statistic: mean ηθ|y
– One-sample t-test: Parameter > 0?
– Paired t-test:
parameter 1 > parameter 2?
– rmANOVA: e.g. in case of multiple
sessions per subject
Inference about DCM parameters
Bayesian parameter averaging
![Page 39: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/39.jpg)
inference on model structure or inference on model parameters?
inference on
individual models or model space partition?
comparison of model
families using
FFX or RFX BMS
optimal model structure assumed to
be identical across subjects?
FFX BMS RFX BMS
yes no
inference on
parameters of an optimal model or parameters of all models?
BMA
definition of model space
FFX analysis of parameter
estimates
(e.g. BPA)
RFX analysis of parameter
estimates
(e.g. t-test, ANOVA)
optimal model structure assumed to
be identical across subjects?
FFX BMS
yes no
RFX BMS
Stephan et al. 2010, NeuroImage
Sequence of analysis
![Page 40: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/40.jpg)
Model comparison and selection
Given competing hypotheses
on structure & functional
mechanisms of a system, which
model is the best?
For which model m does p(y|m)
become maximal?
Which model represents the
best balance between model
fit and model complexity?
Pitt & Miyung (2002) TICS
![Page 41: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/41.jpg)
pmypAIC ),|(log
Logarithm is a
monotonic function
Maximizing log model evidence
= Maximizing model evidence
)(),|(log
)()( )|(log
mcomplexitymyp
mcomplexitymaccuracymyp
In SPM2 & SPM5, interface offers 2 approximations:
Np
mypBIC log2
),|(log
Akaike Information Criterion:
Bayesian Information Criterion:
Log model evidence = balance between fit and complexity
Penny et al. 2004, NeuroImage
Approximations to the model evidence in DCM
No. of
parameters
No. of
data points
AIC favours more complex models,
BIC favours simpler models.
![Page 42: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/42.jpg)
The negative free energy approximation
• The negative free energy F is a lower bound on the log model
evidence:
mypqKLmypF ,|,)|(log
F comprises the expected log likelihood and the Kullback-Leibler (KL) divergence between conditional and prior densities
![Page 43: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/43.jpg)
The complexity term in F
• In contrast to AIC & BIC, the complexity term of the negative free energy F accounts for parameter interdependencies. Under gaussian assumptions:
• The complexity term of F is higher
– the more independent the prior parameters
– the more dependent the posterior parameters
– the more the posterior mean deviates from the prior mean
• NB: SPM8 only uses F for model selection !
y
T
yy CCC
mpqKL
|
1
||2
1ln
2
1ln
2
1
)|(),(
![Page 44: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/44.jpg)
Bayes factors
)|(
)|(
2
112
myp
mypB
positive value, [0;[
For a given dataset, to compare two models, we compare their
evidences.
B12 p(m1|y) Evidence
1 to 3 50-75% weak
3 to 20 75-95% positive
20 to 150 95-99% strong
150 99% Very strong
Kass & Raftery classification:
Kass & Raftery 1995, J. Am. Stat. Assoc.
or their log evidences
2112)ln( FFB
![Page 45: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/45.jpg)
Inference on model space
Model evidence: The optimal balance of fit and complexity
Comparing models
• Which is the best model?
1 2 3 4 5 6 7 8 9 10
model
lme
![Page 46: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/46.jpg)
Inference on model space
Model evidence: The optimal balance of fit and complexity
Comparing models
• Which is the best model?
Comparing families of models
• What type of model is best?
• Feedforward vs feedback
• Parallel vs sequential processing
• With or without modulation
1 2 3 4 5 6 7 8 9 10
model
lme
![Page 47: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/47.jpg)
Model evidence: The optimal balance of fit and complexity
Comparing models
• Which is the best model?
Comparing families of models
• What type of model is best?
• Feedforward vs feedback
• Parallel vs sequential processing
• With or without modulation
Only compare models with the same data
1 2 3 4 5 6 7 8 9 10
model
lme
A
D
B
C
A
B
C
Inference on model space
![Page 48: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/48.jpg)
To recap... so, DCM….
• enables one to infer hidden neuronal processes from fMRI data
• allows one to test mechanistic hypotheses about observed effects
– uses a deterministic differential equation to model neuro-dynamics (represented
by matrices A,B and C).
• is informed by anatomical and physiological principles.
• uses a Bayesian framework to estimate model parameters
• is a generic approach to modelling experimentally perturbed dynamic
systems.
– provides an observation model for neuroimaging data, e.g. fMRI, M/EEG
– DCM is not model or modality specific (Models can change and the method extended to
other modalities e.g. ERPs, LFPs)
![Page 49: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/49.jpg)
Applications & extensions of DCM to fMRI data
Brain connectivity: types & definitions
Anatomical connectivity Functional connectivity Effective connectivity
Overview
Dynamic causal models (DCMs)
Neuronal model Hemodynamic model
Estimation: Bayesian framework
![Page 50: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/50.jpg)
V1
V5
SPC Photic
Motion
Time [s]
Attention
We used this model to assess the site of attention
modulation during visual motion processing in an
fMRI paradigm reported by Büchel & Friston.
Friston et al. 2003,
NeuroImage
Attention to motion in the visual system
- fixation only
- observe static dots + photic V1
- observe moving dots + motion V5
- task on moving dots + attention V5 + parietal cortex
?
![Page 51: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/51.jpg)
V1
V5
SPC
Motion
Photic
Attention
0.85
0.57 -0.02
1.36 0.70
0.84
0.23
Model 1:
attentional modulation
of V1→V5
V1
V5
SPC
Motion
Photic Attention
0.86
0.56 -0.02
1.42
0.55
0.75
0.89
Model 2:
attentional modulation
of SPC→V5
Comparison of two simple models
Bayesian model selection: Model 1 better than model 2
→ Decision for model 1: in this experiment, attention
primarily modulates V1→V5
1 2log ( | ) log ( | )p y m p y m
![Page 52: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/52.jpg)
Planning a DCM-compatible study
• Suitable experimental design:
– any design that is suitable for a GLM
– preferably multi-factorial (e.g. 2 x 2)
• e.g. one factor that varies the driving (sensory) input
• and one factor that varies the contextual input
• Hypothesis and model:
– Define specific a priori hypothesis
– Which parameters are relevant to test this hypothesis?
– If you want to verify that intended model is suitable to test this hypothesis,
then use simulations
– Define criteria for inference
– What are the alternative models to test?
![Page 53: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/53.jpg)
Multifactorial design:
explaining interactions with DCM
Task factor Task A Task B
Stim
1
Stim
2
Stim
ulu
s fa
cto
r
A1 B1
A2 B2
X1 X2
Stim2/
Task A
Stim1/
Task A
Stim 1/
Task B
Stim 2/
Task B
GLM
X1 X2
Stim2
Stim1
Task A Task B
DCM
Let’s assume that an SPM analysis
shows a main effect of stimulus in
X1 and a stimulus task interaction
in X2.
How do we model this using DCM?
![Page 54: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/54.jpg)
X1 X2
Stim2
Stim1
Task A Task B
Stim 1
Task A
Stim 2
Task A
Stim 1
Task B
Stim 2
Task B
Simulated data
+ +++
+
+++
+
+++
X1
X2
![Page 55: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/55.jpg)
Stim 1
Task A
Stim 2
Task A
Stim 1
Task B
Stim 2
Task B
plus added noise (SNR=1)
X1
X2
![Page 56: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/56.jpg)
Slice timing model
• potential timing problem in DCM:
temporal shift between regional
time series because of multi-slice
acquisition
• Solution:
– Modelling of (known) slice timing of each area.
1
2
slic
e a
cquis
itio
n
visual
input
Slice timing extension now allows for any slice timing differences!
Long TRs (> 2 sec) no longer a limitation.
(Kiebel et al., 2007)
![Page 57: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/57.jpg)
Two-state model
)(tu
ijij uBA
input
Single-state DCM
1x
Intrinsic (within-region)
coupling
Extrinsic (between-region)
coupling
NNNN
N
x
x
tx
AA
AA
A
CuxuBAt
x
1
1
111
)(
)(
Two-state DCM
Ex1
)exp( ijij uBA
Ix1
11 11exp( )IE IEA uBIEx ,
1
CuzBuAzzjj
I
N
E
N
I
E
A A
A A A
A A
A A A
u
x
x
x
x
t x
e e
e e e
e e
e e e
A
Cu x AB t
x
II NN
IE NN
EI NN
EE NN N
II IE
N EI EE
1
1
) (
0 0
0
0 0
0
) (
1
11 11
1 11 11
![Page 58: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/58.jpg)
SPCSPC
SPCSPCVSPC
VV
SPCVVVVV
VV
VVVV
IIIE
EIEEEE
IIIE
EEEIEEEE
IIIE
EEEIEE
A
0000
000
0000
00
0000
000
5
55
55515
11
5111
Attention
SPCSPC
SPCSPCVSPC
VV
SPCVVVVV
VV
VVVV
IIIE
EIEEEE
IIIE
EEEIEEEE
IIIE
EEEIEE
A
0000
000
0000
00
0000
000
5
55
55515
11
5111
Attention
SPCSPC
SPCSPCVSPC
VV
SPCVVVVV
VV
VVVV
IIIE
EIEEEE
IIIE
EEEIEEEE
IIIE
EEEIEE
A
0000
000
0000
00
0000
000
5
55
55515
11
5111
Attention
DCM for Büchel & Friston
- FWD
- Intr
- BCW
b
Exa
mp
le:
Tw
o-s
tate
Mo
del C
om
pari
so
n
![Page 59: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/59.jpg)
bilinear DCM
CuxDxBuAdt
dx m
i
n
j
j
j
i
i
1 1
)()(CuxBuA
dt
dx m
i
i
i
1
)(
Bilinear state equation
u1
u2
nonlinear DCM
Nonlinear state equation
u2
u1
Here DCM can model activity-dependent changes in connectivity; how
connections are enabled or gated by activity in one or more areas.
Nonlinear DCM for fMRI
![Page 60: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/60.jpg)
V1 V5
stim
PPC
attention
motion
-2 -1 0 1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
%1.99)|0( 1,5 yDp PPC
VV
1.25
0.13
0.46
0.39
0.26
0.50
0.26
0.10 MAP = 1.25
Stephan et al. 2008, NeuroImage
Can V5 activity during attention to motion be explained by
allowing activity in PPC to modulate the V1-to-V5 connection?
Nonlinear DCM for fMRI
![Page 61: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/61.jpg)
Daunizeau et al, 2009
Friston et al, 2008 Inversion: Generalised filtering (under the Laplace assumption)
activity
x1(t)
activity
x2(t) activity
x3(t)
neuronal
states
t
driving
input u1(t)
modulatory
input u2(t)
t
Stochastic innovations: variance hyperparameter
Stochastic DCMs More recent developments
Although here
driving inputs are
not necessary...
![Page 62: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/62.jpg)
Bayesian Model Selection
for large model spaces
• for less constrained model
spaces, search methods are
needed
• fast model scoring via the
Savage-Dickey density ratio:
2 3 4 5 6 10
0
10 1
10 2
10 3
10 4
10 5
Number of models
number of nodes
-600 -500 -400 -300 -200 -100 0 100 -600
-500
-400
-300
-200
-100
0
100
Log-evidence
Savage-Dickey
Fre
e-en
ergy
1 /22
n n
iA m
assuming reciprocal
connections
ln |
ln 0 | ln 0 |
i
i i i F
p y m
q m p m
Friston et al. 2011, NeuroImage
Friston & Penny 2011, NeuroImage
Life easier for more
exploratory approachs!
More recent developments
![Page 63: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/63.jpg)
• DCM is not one specific model, but a framework for Bayesian
inversion of dynamical systems
• The default implementation in SPM is evolving over time:
- better numerical routines for inversion
- changes in prior to cover new variants
To enable replication of your results, you should state which SPM
version you are using!
The evolution of DCM in SPM
![Page 64: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/64.jpg)
Some introductory references • The first DCM paper: Dynamic Causal Modelling (2003). Friston et al.
NeuroImage 19:1273-1302.
• Physiological validation of DCM for fMRI: Identifying neural drivers with
functional MRI: an electrophysiological validation (2008). David et al. PLoS
Biol. 6 2683–2697
• Hemodynamic model: Comparing hemodynamic models with DCM (2007).
Stephan et al. NeuroImage 38:387-401
• Nonlinear DCMs:Nonlinear Dynamic Causal Models for FMRI (2008). Stephan
et al. NeuroImage 42:649-662
• Two-state model: Dynamic causal modelling for fMRI: A two-state model
(2008). Marreiros et al. NeuroImage 39:269-278
• Group Bayesian model comparison: Bayesian model selection for group
studies (2009). Stephan et al. NeuroImage 46:1004-10174
• 10 Simple Rules for DCM (2010). Stephan et al. NeuroImage 52.
![Page 65: Dynamic causal modelling for fMRI - FIL | UCL€¦ · Dynamic causal modelling for fMRI Friday 18th May 2012 Statistical Parametric Mapping for fMRI ... Basics of Dynamic Causal Modelling](https://reader036.vdocuments.us/reader036/viewer/2022070716/5ed99fe15139c40fce67556e/html5/thumbnails/65.jpg)
Thank you for your attention!!!