4. Experimental Design and Analysis in fMRI

Question 1

BOLD (___) signal is an ___ measure of ____

Answer 1

Blood oxygenation level dependent signal is an indirect measure of neural activity

Question 2

What is cognitive subtraction (fRMI design)?

Answer 2

Choosing two tasks that only differ in one critical respect, then compare the BOLD signals generated by each task

• regions showing a difference in avg activity may be contributing to the varied element of the task

Question 3

Which experimental design should fMRI experiments use?

Answer 3

Because of individual variation in the BOLD response and anatomy, fMRI experiments should use a WITHIN-SUBJECTS design wherever possible

Question 4

Why is a block design very important for investigating activity with fMRI?

Answer 4

The responses to individual trials cannot be distinguished, but add-up to produce a robust signal change.

Question 5

What is the differences between longer and shorter blocks in fMRI experiment designs?

Answer 5

• At longer interblock intervals, the hemodynamic response can return to baseline during the control condition
• Shorter blocks allow more repetitions, but the sluggish hemodynamic response means less signal change – harder to distinguish from noise.

Question 6

What are the advantages and disadvantages of a block design in fMRI experiments?

Answer 6

+ excellent power to detect active voxels; improved signal to noise (relative to event related design)

• responses to individual trials cannot be analyzed

Question 7

What is an Event-related design in fMRI experiments?

Answer 7

• Each event (typically a single experimental trial or part of it) is modelled separately.
• Trials are carefully spaced and ordered so that time-locked responses can be estimated for different classes of trial

Question 8

What are the advantages and disadvantages of event-related design in fMRI?

Answer 8

+ data can be analyzed based on responses (e.g. sorting trials into correct and incorrect and comparing them)

+ Timecourse of the BOLD response can be accurately estimated

• power to detect active voxels is reduced

Question 9

fMRI preprocessing: Motion correction

Answer 9

Successive images are aligned with one-another

Question 10

fMRI preprocessing: Spatial smoothing

Answer 10

Blurring (increases signal to noise)

Question 11

fMRI preprocessing: High pass temporal filtering removes

Answer 11

Gradual changes in signal intensity

Question 12

What do statistical maps indicate? (fMRI)

Answer 12

maps indicate the probability of a type 1 error (false positive), not absolute signal change

• the lower this probability (p-value) the brighter/hotter the colour

Question 13

What is the Bonferroni correction?

Answer 13

takes account of the number of comparisons made (too conservative for fMRI)

Question 14

What is the FWE (Family Wise Error) voxel correction?

Answer 14

Takes account of number of INDEPENDENT comparisons

Question 15

What is the FDR (False discovery rate) correction?

Answer 15

Controls false-discovery rate

Question 16

What is the cluster-size correction? (fMRI)

Answer 16

Takes account of contiguity (the state of bordering or being in contact with something) of active voxels

Question 17

What are whole brain analysis useful for?

Answer 17

which brain regions are selectively engaged in a given task:

• spatial pattern of activity associated with different tasks: gives an indication of which tasks overlap in terms of the underlying brain processes.
• identify range of tasks which activate specific brain regions: gives an indication of what specific brain regions do.

Question 18

One way to avoid the problem of multiple comparisons? (fMRI)

Answer 18

Restricting analysis to data from a particular part of the brain, IDENTIFIED in ADVANCE

Question 19

ROIs can be defined using ____ or ____ criteria

Answer 19

anatomical or functional

Question 20

What are the advantages and disadvantages of ROI analysis?

Answer 20

+ Increased statistical power

+ more targeted approach - leads to better specified hypotheses?

+ can be combined with whole brain analysis

• misses potentially important patterns outside the ROI
• danger of biased post-hoc selection of ROI

Question 21

stats in a combined statistical map take into account ____ ____

Answer 21

individual variation

Question 22

Whole-brain group analysis demands that different brains…

Answer 22

are aligned with one another in the same space

Question 23

fMRI design needs to consider:

Answer 23

1. Timecourse (event related vs block design)

2. Intersubject variability of the hemodynamic response (Within vs between subjects)

Question 24

After spatial and temporal preprocessing, fMRI analysis uses a ____ to estimate task related changes in the BOLD signal

Answer 24

general linear model (GLM)

Question 25

(In fMRI) The statistical significance of differences between conditions is determined by…

Answer 25

mapping statistical parameters to brain voxels, using appropriate thresholds to avoid false positives (Type 1 errors)

Question 26

To support robust statistical inferences which will/can be extended to a wider population, multi-subject fMRI experiments typically employ:

Answer 26

Random/mixed-effects, within-subjects design

Question 27

In the analysis of a typical fMRI experiment, BOLD signal changes are modelled by combining predicted timeseries for each distinct class of event (or block) which are fitted to the fMRI data…

Answer 27

to minimize residual variation

Question 28

If BOLD responses to individual experimental trials of the same task are to be compared (e.g., correct versus incorrect trials) in an fMRI study, it is essential to use:

Answer 28

an event related design