I did not have a specific application in mind when I wrote it. Part of it was to help look for artifacts (ie, structured noise in the residuals). It will always give you a number of eigenvectors equal to the smaller of the number of "temporal" dimensions or spatial dimensions. For imaging, this is usually the "temporal" (14 in your case).
doug On 12/05/2014 12:00 PM, Thomas DeRamus wrote: > Appreciate the quick reply. I may be re-framing my question a bit > here, but I'm curious as to the context in which one would use the PCA > analysis. Is it meant primarily for functional data or can it be > applied to glmfits for structural data? When I ran this the first > time, I got 14 eigenvalues on a subset of 14 participant MPRAGE's from > a large dataset to test the --pca flag (I haven't seen much about it > on the documentation). In hindsight, I think the 14 eigenvalues I'm > getting are actually the number of participants in the study, which > means I am probably using this flag in the wrong context. I'm curious > about using this on a larger dataset from different types of scanners > and finding components unique to scanner type using pca that can be > addressed in later group analyses. Given that PCA uses "temporal" > components (if you don't mind could you expand on this?) can it even > be applied to morphological data like individual T1 images? Or should > it be used solely in isolation with functional images for individual > subjects (such as the way GIFT or ICA toolboxes would?). > > Basically I'd like to know conceptually when you would use this and > what it is capable of doing. My apologies if that is a horrible > combination of vague and broad. > > On Thu, Dec 4, 2014 at 1:50 PM, Douglas N Greve > <[email protected] <mailto:[email protected]>> wrote: > > > It performs pca/svd on the residuals. The sdiag.mat I think > includes the > singular values but also the percent variance explained by each > component and the cumulative explained by all components upto and > including that component. You probably just need to change the overlay > thresholds on the v.mgh to make it look more reasonable. The range of > the singular values depends on your data so you can't just look at > them > and decide they are too big. The u.mtx are the "temporal" components. > > doug > > On 12/04/2014 11:10 AM, Thomas DeRamus wrote: > > Dear Freesurfer experts, > > > > Potentially super-noob question here, but I'm curious about the > --pca > > flag on mri_glmfit. I can see that it does PCA/SVD on the > dataset? in > > question and saves the residuals to the glmdir specified, but I was > > unable to find anything in the documentation describing what exactly > > it is doing. I got sdiag.mat, stats.dat (with some rather large > > eigenvalues, on the order of 40k, but they might look more > normal once > > I demean my variables), a u.mtx, a v.mgh surface overlay, and an > error > > log (stating MatlabRead: readHeder returned NULL, ImageRead > > (/path/glmdir/pca-eres/sdiag.mat) failed. When I loaded the > v.mgh, it > > looked rather binarized (with red and blue colors covering the > entire > > cortex mask for that hemi). > > > > May I ask how it is computing these eigenvalues? What the sdiag and > > u.mtx files are used for? And what is being displayed in the > v.mgh (is > > it components? if so, what do these components reflect?). And most > > importantly, is this meant primarily for functional data or can > it be > > applied to mophometric data? If it can be used for the latter, > what is > > being fed into the PCA/SVD to determine spatial components? > > > > -- > > *Thomas DeRamus* > > UAB Department of Psychology, Behavioral Neuroscience > > Graduate Research Trainee > > Civitan International Research Center > > 1719 6th Ave S, Suite 235J, Birmingham, AL 35233 > > _Phone_: 205-934-0971 <tel:205-934-0971> _Email:_ > [email protected] <mailto:[email protected]> > > <mailto:[email protected] <mailto:[email protected]>>, > [email protected] <mailto:[email protected]> <mailto:[email protected] > <mailto:[email protected]>> > > > > > > _______________________________________________ > > Freesurfer mailing list > > [email protected] > <mailto:[email protected]> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > -- > Douglas N. Greve, Ph.D. > MGH-NMR Center > [email protected] <mailto:[email protected]> > Phone Number: 617-724-2358 <tel:617-724-2358> > Fax: 617-726-7422 <tel:617-726-7422> > > Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting > <http://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting> > FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 > www.nmr.mgh.harvard.edu/facility/filedrop/index.html > <http://www.nmr.mgh.harvard.edu/facility/filedrop/index.html> > Outgoing: > ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/ > > _______________________________________________ > Freesurfer mailing list > [email protected] <mailto:[email protected]> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > The information in this e-mail is intended only for the person to > whom it is > addressed. If you believe this e-mail was sent to you in error and > the e-mail > contains patient information, please contact the Partners > Compliance HelpLine at > http://www.partners.org/complianceline . If the e-mail was sent to > you in error > but does not contain patient information, please contact the > sender and properly > dispose of the e-mail. > > > > > -- > *Thomas DeRamus* > UAB Department of Psychology, Behavioral Neuroscience > Graduate Research Trainee > Civitan International Research Center > 1719 6th Ave S, Suite 235J, Birmingham, AL 35233 > _Phone_: 205-934-0971 _Email:_ [email protected] > <mailto:[email protected]>, [email protected] <mailto:[email protected]> > > > _______________________________________________ > Freesurfer mailing list > [email protected] > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer -- Douglas N. Greve, Ph.D. MGH-NMR Center [email protected] Phone Number: 617-724-2358 Fax: 617-726-7422 Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/ _______________________________________________ Freesurfer mailing list [email protected] https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
