Upload
nicola
View
78
Download
0
Embed Size (px)
DESCRIPTION
Dynamic Causal Modelling for evoked responses . Stefan Kiebel. Max Planck Institute for Human Cognitive and Brain Sciences Leipzig, Germany. Overview of the talk. 1 M/EEG analysis 2 Dynamic Causal Modelling – Motivation 3 Dynamic Causal Modelling – Generative model - PowerPoint PPT Presentation
Citation preview
Dynamic Causal Modellingfor evoked responses
Stefan Kiebel
Max Planck Institute forHuman Cognitive and Brain Sciences
Leipzig, Germany
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
pseudo-random auditory sequence
80% standard tones – 500 Hz
20% deviant tones – 550 Hz
time
standards deviants
Mismatch negativity (MMN)
time (ms)
μV
Paradigm
Raw data(e.g., 128 sensors)
Preprocessing (SPM8)
Evoked responses(here: single sensor)
Electroencephalography (EEG)
time
sens
ors
sens
ors
standard
deviant
time (ms)
amplitude (μV)
Analysis at sensor levelse
nsor
sse
nsor
s
standard
deviant
time
Conventional approach: Reduce evoked response to a few
variables.
Alternative approach?
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
Electroencephalography (EEG)
time (ms)
amplitude (μV)
Modelling aim: Explain all data with few
parameters
How?Assume data are caused
by few communicating brain sources
Connectivity models
A1 A1
STG
Input (stimulus)
STG
Conventional analysis: Which regions are involved in task?
A1 A1
STG
Input (stimulus)
STG
DCM analysis: How do regions communicate?
Macro- and meso-scale
internal granularlayer
internal pyramidallayer
external pyramidallayer
external granularlayer
AP generation zone synapses
macro-scale meso-scale micro-scale
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
The generative model
),,( uxfx
Source dynamics
states x parameters θ
Input u
Evoked response
data y
),( xgy
Spatial forward model g
David et al., NeuroImage, 2006Kiebel et al., Human Brain Mapping, 2009
Neural mass equations and connectivity
Extrinsicforward
connectionsspiny
stellate cells
inhibitory interneurons
pyramidal cells
4 3
214
014
41
2))()((ee
LF
e
e xxCuxSIAAHx
xx
1 2)( 0xSAF
)( 0xSAL
)( 0xSABExtrinsic backward connections
Intrinsic connections
neuronal (source) model
Extrinsic lateral connections
State equations
,,uxfx
0x
278
038
87
2))()((ee
LB
e
e xxxSIAAHx
xx
236
746
63
225
1205
52
650
2)(
2))()()((
iii
i
ee
LB
e
e
xxxSHx
xx
xxxSxSAAHx
xxxxx
David et al., NeuroImage, 2006Time (ms)
Am
plitu
de (a
.u.)
Model for auditory evoked response
Garrido et al., PNAS, 2007
Spatial model
0x
LL
Depolarisation ofpyramidal cells
Spatial model
Sensor data y
Kiebel et al., NeuroImage, 2006Daunizeau et al., NeuroImage, 2009
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
Bayesian model inversion
Evoked responsesSpecify generative forward model
(with prior distributions of parameters)
Expectation-Maximization algorithm
Iterative procedure: 1. Compute model response using current set of parameters
2. Compare model response with data3. Improve parameters, if possible
1. Posterior distributions of parameters
2. Model evidence )|( myp
),|( myp
Friston, PLoS Comp Biol, 2008
Model selection: Which model is the best?
)|( 1mypModel 1
data y
Model 2
...
Model n
)|( 2myp
)|( nmypbest?
Model selection:
Selectmodel with
highestmodel
evidence
),|( 1myp
),|( 2myp
),|( nmyp
)|( imyp
Fastenrath et al., NeuroImage, 2009Stephan et al., NeuroImage, 2009
best?
Overview of the talk
1 M/EEG analysis
2 Dynamic Causal Modelling – Motivation
3 Dynamic Causal Modelling – Generative model
4 Bayesian model inversion
5 Examples
Auditory evoked response
Garrido et al., PNAS, 2007
Auditory evoked response
Garrido et al., PNAS, 2007
time (ms) time (ms)
A1 A1
STG STG
ForwardBackward
Lateral
STG
input
A1 A1
STG STG
ForwardBackward
Lateral
input
A1 A1
STG
ForwardBackward
Lateral
input
Forward - F Backward - BForward and
Backward - FB
STG
IFGIFGIFG
modulation of effective connectivity
Mismatch negativity
Garrido et al., (2007), NeuroImage
Bayesian Model Comparison
Forward (F)
Backward (B)
Forward and Backward (FB)
subjects
log
-evi
denc
e
Group level
Group model comparison
Garrido et al., (2007), NeuroImage
Summary
DCM enables testing hypotheses about how brain sources communicate.
DCM is based on a neurobiologically grounded, dynamic model of evoked responses.
Differences between conditions are modelled as modulation of connectivity.
Inference: Bayesian model selection
Thanks to: Marta GarridoJean DaunizeauKarl Friston
and the FIL methods group