I dug a little more and found out more information. The 64 input images of
16 subjects (4 cope images / subject) are from two different lower level
analysis approaches. cope images from 7 of 16 subjects are from the first
level analysis from individual subjects whereas cope images from the other 9
subjects are from the second level analysis pooling over different runs. But
there are of the same nature/condition (from different design, long story).
Therefore when generating tdof images at the higher level ANOVA analysis,
FEAT only found 36 (9x4) tdof images from the second level analysis from the
9 subjects. So the following flame command failed with the error message
about the "index=37".
/usr/local/fsl/bin/flame --cope=tmpcope --vc=tmpvarcope --dvc=tmptdof
--mask=tmpmask --ld=stats0 --dm=design.mat --cs=design.grp --tc=desig
n.con --ols --nj=10000 --bi=500 --se=1 --fm --zlt=2.25 --zut=2.65
After getting rid of --dvc=tmptdof, I can continue running the flame process
manually.
My questions are, then, is this approach still valid? What function do the
tdof images serve for higher level analysis involving more than 2 stages
(e.g. subject level, run/session level, group level), since there are no
such images for the 2nd level group analysis?
If it is valid, how to avoid doing this manually by re-run the command
logged in the report.com file? Which source scripts should be modified to
prevent using --dvc options and avoid errors when using mixture of cope
images from the first (subject) and second (run/session) levels?
Thanks for any suggestion.
|