Canonical PTA3 modes on fMRI data
Now an example with fMRI data (brain x time x subjects) for a multi-subject verbal experiment.(To perform a multi-subject analysis coregistration of the 3D volumes are performed using motion correction and registration to a standard.) The first analysis is without metrics or penalisations (canonical). The time series (for each voxel and eac subject) were detrended, and each subject was scaled to variance one. (here the brain is only one slice: the label on the brain component is wrong slices all the 14:18 were not used.)
Each set of triple of components are viewed together and interpreted simultaneously.
The first Principal Tensor i.e. singular value vs111 (figure of the triple of components):
The experiment is meant to show verbal activation (left fronto-temporal on te right of the image) following the verbal paradigm (the blue square wave). There should not be any differences in the sex differenciated groups (all right handed) and there is not -Wilcoxon p=0.17- but if you test without the "green" Male -Wilcoxon p=0.03-. (Removing this subject before analysis didn't change the other components and the subject component was still dispersed for the male.
back to front-page
Last modified: Thu Jul 5 10:22:41 BST 2001