Dear Kayako,
|Question 1,
|I acquired a blocked-designed fMRI dataset which comprises three task
|conditions, A, B and C, repeated in this order in a session. The dataset
|has no rest or control conditions. I would like to know the brain regions
|which are responsible for each of the three tasks (that is, regions for
|A, regions for B and regions for C). I suppose one of the following
|procedures would do for this purpose.
|
|(1) In the 'fMRI models' specification, 'number of conditions or trials'
|should be specified as "3". In the 'Select contrasts...' section of the
|'Results' specification, contrasts should be defined as "1 0 0" for task
|A, "0 1 0" for task B and "0 0 1" for task C.
|
|(2) Statistics should be performed separately for each of the three
|tasks. In the 'fMRI models' specification, 'number of conditions or
|trials' should be specified as "1" for a task and the other conditions
|should not be mentioned. In the 'Select contrasts...' section of the
|'Results' specification, a contrast should be defined as "1". This
|procedure would be repeated for three times, for task A, B and C.
|
|Which, or the other, procedure is appropriate ?
The first of these two. In general, it is best to model all
experimental conditions; failure to do so may (will) result in the
residuals containing systematic (unmodelled) structure which is not a good
thing.
|Question 2,
|When analyzing fMRI datasets which comprise multiple task conditions (A,
|B, C for example) and a control (rest) condition (R), which of the below,
|or the other, procedures is followed in SPM99b ?
|
|(1) In SPM99b, the control (rest) condition is unspecified by users.
|Instead, 'SOA' and 'epoch length' for the control are calculated by those
|inputted for tasks. Thus, contrasts such as 'A - R', 'B - R' and 'C - R'
|can be defined in addition to 'A - B', 'B - C' and so on.
|
|(2) When analyzing for task A, the other tasks (B, C) and rest (R) are
|handled together as 'control', compared with task A. Thus, activation
|maps are generated by contrasts such as 'A - BCR', 'B - ACR' or 'C - ABR'.
If I've understood your question, the first of these procedures is
an appropriate characterisation. Again, as a general rule the activation
produced by all tasks should be modelled, even if you are only interested
in (say) task A.
best wishes,
Geraint
-----------------------------------------------------------------
Dr. Geraint Rees
Wellcome Advanced Fellow Lecturer
California Institute of Technology Institute of Neurology
Division of Biology 139-74 University College London
Pasadena 12 Queen Square
California 91125 London WC1N 3BG
voice (626) 395-2880 voice (171) 833-7472
fax (626) 796-8876 fax (171) 813-1420
http://www.klab.caltech.edu/~geraint [log in to unmask]
-----------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|