Thank you Steve,
actually I was thinking to do an SVD of all subjects' timecourses for a given IC (using fslmeants --eig), exactly because of the phase difference. I was thinking this because of a note in Jill O'Reilly's 2010 paper on the Cerebellum, where the eigen timecourse is proposed as an alternative to the mean time course (of a certain ROI) exactly because of the problem of phase difference: "As a Gedankenspiel, imagine 2 perfectly anticorrelated voxels of identical variance: in this case the mean would be constant and therefore fail to characterize the observed variations, whereas the Eigen time series fully describes the signal dynamics." (p.957)
> You mean (for a given IC) feed all subjects' timecourses into PCA?
> That doesn't make sense as there is no consistent phase relationship
> between subjects - you will probably end up with roughly the right PS,
> but will take a bad SNR hit (compared with the above approach) because
> the subjects' different timecourses will "average out".
all the best!
leonardo
--
Leonardo Cerliani, PhD
BCN Neuro-Imaging Center
University Medical Center Groningen
Antonius Deusinglaan 2 (room 131)
tel. +31.363.8805
9713 AW Groningen - The Netherlands
Social Brain Lab
Netherlands Institute for Neuroscience
Royal Netherlands Academy of Arts and Sciences
Meibergdreef 47, 1105 BA Amsterdam, NL
___
{o,o}
|)__)
--"-"--
http://openlibrary.org/people/leonardocerliani/lists
|