## Presentation on theme: "Event-related fMRI"— Presentation transcript

Slide1

Event-related fMRI

SPM course May 2015Helen BarronWellcome Trust Centre for Neuroimaging12 Queen SquareSlide2

Overview

Event-related design vs block designModelling eventsOptimising the designSlide3

Overview

Event-related design vs block designModelling eventsOptimising the designSlide4

Center for Vital Longevity Face Database

Berkeley Segmentation Datasettime

faces

scenes

faces

scenes

S

cenes vs face processing

How should we order the presentation of the stimuli?

What timing should we use between presentations? Slide5

Brief

StimulusUndershoot

Initial

Undershoot

Peak

BOLD

r

esponse

Initial undershoot

Peak 4-6s post-stimulus

Undershoot before returning to baseline

This is SLOW.

How should we present our stimuli?Slide6

Experimental Designs

time

scene

face

Intermixed / Event-Related Design

time

Block / Epoch Design

activity in face area

activity in scene area

Blocked designs have high statistical power so why would we want to use event-related design?

The order is randomSlide7

Event-Related Designs

Advantages over block designs:Post-hoc classification of trials by the experimentere.g. by subsequent memory, Wagner et al., 1998750ms

cheese

+

1250ms

+

2000ms

time

Word trial (2

secs

)

Fixation trial (2

secs

)

“Null event”Slide8

Event-Related Designs

Advantages over block designs:Events which can only be indicated by the participante.g. decision making, perceptual changes, Kleinschmidt et al., 1998Slide9

Event-Related Designs

Advantages over block designs:Paradigms which cannot be blockedwhere surprise is important, oddball designs

timeSlide10

Event-Related Designs

Advantages over block designs:Post-hoc classification of trials by the experimentere.g. by subsequent memoryEvents which can only be indicated by the participante.g. decision-making , perceptual changesParadigms that cannot be blocked

e.g. oddball designsSlide11

Overview

Event-related design vs block designModelling eventsOptimising the designSlide12

Modelling events

Model

Block / Epoch

X

Design

time

time

Model

Design

Event relatedSlide13

ITI(Inter-Trial Interval)

TrialTrialTrial

SOA

(Stimulus Offset Asynchrony)

Trial + ITI

ITI

(Inter-Trial Interval)

time

Terminology

for consistency with previous literature

Event: brief stimulus presentation thought to lead to a brief burst in neural activity

Epoch: sustained stimulus presentation thought to lead to sustained neural activity

Impulse response: BOLD response to an eventSlide14

To infer the contribution of a given voxel to house or scene processing we need to model the events in a design matrix

=

+

X

The GLMSlide15

The design matrix

Regressor

1: Face

Regressor

2: Scene

Regressor

3: Constant

X=

Brief

Stimulus

Undershoot

Initial

Undershoot

Peak

We need to model the impulse response functionSlide16

time

Design matrix

convolution

down-sample for each scan

Temporal basis

functions

Events across time

The design matrixSlide17

Finite Impulse Response (FIR)

Fourier

Temporal basis function

Gamma functionSlide18

Canonical

Haemodynamic Response Function (HRF) used in SPM2 gamma functionsAssumed to be the same everywhere in the brain

Canonical

Temporal basis functions

the standard HRFSlide19

Canonical

Haemodynamic Response Function (HRF) used in SPM2 gamma functions+Multivariate Taylor expansion in time (Temporal Derivative)Canonical

Temporal

Negatively weight

temporal

Positively weight temporal

Temporal basis functions

the standard HRF and derivativesSlide20

Canonical

Haemodynamic Response Function (HRF) used in SPM2 gamma functions+Multivariate Taylor expansion in time (Temporal Derivative)+ Multivariate Taylor expansion in width (Dispersion Derivative)

Canonical

Temporal

Dispersion

Temporal basis functions

the standard HRF and derivatives

Now it is possible to account for variation between brain regionsSlide21

Simple convolution

Which design is more efficient?Slide22

Overview

Event-related design vs block designModelling eventsOptimising the designSlide23

Optimising design: The Aim

We want to:Maximize our t-statistic where there’s an effect – i.e. our efficiency or sensitivityWe need to choose a good:Stimulus orderITISOASlide24

Which SOA is optimal?

Not very efficient…

Very inefficient…

Which design is more efficient?

Neither are very good

16s SOA

4s SOASlide25

Short randomised SOA

Stimulus (“Neural”)HRF

Predicted Data

More efficient…

=

Null eventsSlide26

Block design SOA

Stimulus (“Neural”)HRFPredicted Data

Even more efficient…

=Slide27

=

=

Stimulus (“Neural”)

HRF

Predicted Data

Fourier Transform

Fourier Transform

Block Design, blocks (epochs) = 20s, short ISI

Analysing

efficiency: Fourier transformSlide28

Randomised Design, SOAmin

= 4s, highpass filter = 1/120sFourier TransformFourier Transform

=

=

Stimulus (“Neural”)

HRF

Predicted Data

Analysing

efficiency: Fourier transformSlide29

Sinusoidal modulation, f=1/33s

Fourier TransformFourier Transform=

Stimulus (“Neural”)

HRF

Predicted Data

=

=

The optimal SOASlide30

X: design matrix

c: contrast vector

β

: beta vector

Maximise t by minimising the squared variance

Assuming

σ

is independent of our design, taking a fixed contrast we can only alter our design matrix

Analysing

efficiency: maximising t valueSlide31

Happy (A) vs sad (B) faces: need to know both (A-B) and (A + B)

AB A0.5

0.5

B

0.5

0.5

Transition matrix

Values are probabilities of that condition occurring

Efficiency Example #

1

Two event types, A and B

Randomly intermixed (event-related):

ABBAABABB…

Question: What’s the best SOA to use?

Optimising the SOASlide32

Contrast for Differential

Effect (A-B)

Contrast for Common

Effect (A+B)

SOA (s)

Efficiency

Efficiency Example #1

Note: the optimal SOA for the two contrasts differ

O

ptimal

efficiency

A+B: 16-20s, A-B:

0s

Given a particular design matrix, the different contrasts have different efficiencies.Slide33

Two event types, A and B

Randomly intermixed (event-related) with null events:AB-BAA--B---ABB…

Question: What’s the best SOA to use?

A

B

A

0.33

0.33

B

0.33

0.33

Transition matrix

Values are probabilities of that condition occurring

Efficiency Example #2Slide34

Efficiency Example #2

(

A+B)

(A-B)

SOA (s)

Efficiency

With the addition of null events the optimal SOA

i

s roughly matched for the two contrasts.

O

ptimal

efficiency

A+B:

0s

, A-B:

0s

Should we just use SOAs of 0s?Slide35

Non-linear effects

If the IRs sum in a linear manner then we are OK! But at short SOAs we get non-linearities in the data (saturation effects). Assume linear summation of BOLD response, up to a certain temporal proximity of event

Linear model

Non linear data

(‘saturation effect’)

Linear model is good until SOAs of <1s-2s

Trade off between packing more events

in and having nonlinear saturation effects

which are not modelled.

Friston

et al., 1999Slide36

Efficiency Summary

Block designs:Generally efficient but often not appropriate.Optimal block length 16s with short SOA (beware of high-pass filter).Event-related designs:Efficiency depends on the contrast of interestWith short SOAs ‘null events’ (jittered ITI) can optimise efficiency across multiple contrasts. Non-linear effects start to become problematic at SOA<2sSlide37

Summary

Choosing whether to use an event-related or block designChoosing how to model the BOLD responseOptimising the timing of the experiment (design efficiency)Slide38

Further ReadingBooks

(http://www.fil.ion.ucl.ac.uk/spm/doc/)Statistical Parametric MappingHuman Brain FunctionOnline lecturesSPM Course http://www.fil.ion.ucl.ac.uk/spm/course/video/ Websiteshttp://mindhive.mit.edu/imaging