GUIDE to

The…

D U MM I

E

S’

DCM

Velia Cardin

Functional Specialization is a question of Where?

•Where in the brain is a certain cognitive/perceptual attribute processed?

•What are the Regionally specific effects

your normal SPM analysis (GLM)

Functional Integration is a question of HOW

Experimentally designed input

How does the system work?

What are the inter-regional What are the inter-regional effects?effects?

HowHow do the components of the do the components of the system interact with each system interact with each other?other?

DCM overview•DCM allows you model brain activity at the neuronal level (which is not directly accessible in fMRI) taking into account the anatomical architecture of the system and the interactions within that architecture under different conditions of stimulus input and context.

•The modelled neuronal dynamics (z) are transformed into area-specific BOLD signals (y) by a hemodynamic forward model (λ).

The aim of DCM is to estimate parameters at the neuronal level so that the modelled BOLD signals are most similar to the experimentally measured BOLD signals.

The DCM cycle

Design a study thatallows to investigatethat system

Extraction of time seriesfrom SPMs

Parameter estimationfor all DCMs considered

Bayesian modelselection of optimal DCM

Statistical test on

parameters of optimal

model

Hypothesis abouta neural system

Definition of DCMs as systemmodels

Data acquisition

Planning a DCM-compatible study• Suitable experimental design:

– 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?– ensure that intended model is suitable to test this hypothesis → simulations before

experiment– define criteria for inference

Timing problems at long TRs

• Two potential timing problems in DCM:

1. wrong timing of inputs2. temporal shift between

regional time series because of multi-slice acquisition• DCM is robust against timing errors up to approx. ± 1 s

– compensatory changes of σ and θh

• Possible corrections:– slice-timing (not for long TRs)– restriction of the model to neighbouring regions– in both cases: adjust temporal reference bin in SPM

defaults (defaults.stats.fmri.t0)

1

2

slic

e a

cquis

itio

n

visualinput

Parietal areasV5

Hypothesis A

attention modulates V5 directly

V1

Hypothesis B

Attention modulates effective connectivity between PPC to V5

Defining your hypothesis

+

When attending to motion…….

+

Parietal areasV5

Direct influence

V1

Pulvinar

Indirect influence

DCM cannot distinguish between direct and indirect!

Hypotheses of this nature cannot be tested

Evaluate whether DCM can answer your question

Can DCM distinguish between your hypotheses?

In case of

Practical steps of a DCM study - I

1. Definition of the hypothesis & the model (on paper!)• Structure: which areas, connections and inputs?• Which parameters represent my hypothesis?• How can I demonstrate the specificity of my results?• What are the alternative models to test?

2. Defining criteria for inference:• single-subject analysis: stat. threshold? contrast?• group analysis: which 2nd-level model?

3. Conventional SPM analysis (subject-specific)• DCMs are fitted separately for each session

→ for multi-session experiments, consider concatenation of sessions or adequate 2nd level analysis

Practical steps of a DCM study - II

4. Extraction of time series, e.g. via VOI tool in SPM• cave: anatomical & functional standardisation

important for group analyses!

5. Possibly definition of a new design matrix, if the “normal” design matrix does not represent the inputs appropriately.• NB: DCM only reads timing information of each input

from the design matrix, no parameter estimation necessary.

6. Definition of model• via DCM-GUI or directly

in MATLAB

7. DCM parameter estimation• cave: models with many regions & scans can crash

MATLAB!

8. Model comparison and selection:• Which of all models considered is the optimal one?

Bayesian model selection

9. Testing the hypothesis Statistical test onthe relevant parametersof the optimal model

Practical steps of a DCM study - III

Stimuli 250 radially moving dots at 4.7 degrees/s

Pre-Scanning 5 x 30s trials with 5 speed changes (reducing to 1%)Task - detect change in radial velocity

Scanning (no speed changes)6 normal subjects, 4 x 100 scan sessions;each session comprising 10 scans of 4 different conditions

F A F N F A F N S .................

F - fixation point onlyA - motion stimuli with attention (detect changes)N - motion stimuli without attentionS - no motion

Büchel & Friston 1997, Cereb. CortexBüchel et al. 1998, Brain

V5+

PPCV3A

Attention – No attention

Attention to motion in the visual system

Specify design matrix

• Normal SPM regressors

-no motion, no attention

-motion, no attention

-no motion, attention

-motion, attention

• DCM analysis regressors

-no motion (photic)

-motion

-attention

Defining VOIs

• Single subject: choose co-ordinates from appropriate contrast.

e.g. V5 from motion vs. no motion

• RFX: DCM performed at 1st level, but define group maximum for area of interest, then in single subject find nearest local maximum to this using the same contrast and a liberal threshold (e.g. P<0.05, uncorrected).

DCM button

‘specify’NB: in order!

Can select:

-effects of each condition

-intrinsic connections

-contrast of connections

Input (C)

Output

Latent (intrinsic) connectivity (A)

Modulation of connections (B) Photic

Attention

Motion

V1

IFG

V5

SPC

Motion

Photic

Attention

0.88

0.48

0.37

0.72

0.42

0.66

0.56

0.55•Visual inputs drive V1, Visual inputs drive V1,

activity then spreads to activity then spreads to hierarchically arranged hierarchically arranged visual areas.visual areas.

•Motion modulates the Motion modulates the strength of the V1→V5 strength of the V1→V5 forward connection.forward connection.

•The intrinsic connection The intrinsic connection V1→V5 is insignificant in V1→V5 is insignificant in the absence of motion the absence of motion (a(a2121=-0.05).=-0.05).

•Attention increases the Attention increases the backward-connections backward-connections IFGIFG→SPC and SPC→V5. →SPC and SPC→V5.

A simple DCM of the visual system

0.26

-0.05

Re-analysis of data fromFriston et al., NeuroImage 2003

V1

V5

SPC

Motion

Photic

Attention

0.85

0.57 -0.02

1.360.70

0.84

0.23

V1

V5

SPC

Motion

PhoticAttention

0.86

0.56 -0.02

1.42

0.550.75

0.89V1

V5

SPC

Motion

PhoticAttention

0.85

0.57 -0.02

1.36

0.030.70

0.85

Attention0.23

Model 1:attentional modulationof V1→V5

Model 2:attentional modulationof SPC→V5

Model 3:attentional modulationof V1→V5 and SPC→V5

Comparison of three simple models

Bayesian model selection: Model 1 better than model 2,

model 1 and model 3 equal

→ Decision for model 1: in this experiment, attention

primarily modulates V1→V5

Penny et al. 2004, NeuroImage

Bayes Information Criterion (BIC) and Akaike’s Information Criterion (AIC)

BIC is biased towards simple models

AIC is biased towards complex ones

Make a decision if both factors are in agreement, in particular if

Both provide factors of at least e (2.7183)

DCM button

‘compare’

The read-out in MatLab indicates which model is most likely

–DCM is not exploratory!

DCM is tricky….. ASK the experts!!!

Thanks to Klaas, Ollie and Barrie