Dear Bo,
>It sounds as if there may be a few causes for not finding activation at
>the group level, none of which are related to the specific shape of the
>HRF being used.
>
>First, with only 5 subjects you'll rarely find any activation with random
>effect analyses. You simply don't have the power.
>
>Second, if the individuals have activations in different areas, and there
>is little-to-no common activation across individuals, then nothing is
>going to be significant at the group level regardless of how many subjects
>you have.
>
>Finally, if you're using cluster statistics, then lowering the z-threshold
>doesn't change the leniency at all -- in fact, it's only likely to hide
>smaller (in space) activation regardless of how strong (in Z--scores) they
>are. See previously postings on this list about cluster stats and/or try
>looking at Matthew Brett's except pages on gaussian random field theory at
>http://www.mrc-cbu.cam.ac.uk/Imaging.
Good luck.
Joe
--------------------
Joseph T. Devlin, Ph. D.
FMRIB Centre, Dept. of Clinical Neurology
University of Oxford
John Radcliffe Hospital
Headley Way, Headington
Oxford OX3 9DU
Phone: 01865 222 738
Email: [log in to unmask]
|