 |
 |
Dear Charline, In general in beamforming you might run into more trouble because of using too short window and not having enough data than by making the window too long. But perhaps a justified choice of something in the middle would be the best option. See http://www.ncbi.nlm.nih.gov/pubmed/18155612 for in depth discussion of these issues. Best, Vladimir On Thu, Jan 9, 2014 at 3:56 PM, Urbain Charline <[log in to unmask]> wrote: > Dear Vladimir, > > I am running a MEG study using vectorial beamformer source reconstruction. > As it is often advise to reconstruct sources on smaller time windows (e.g. > 350-400msec) instead of on the whole epoch, I was wondering if, in this > case, we also should compute the covariance matrix on the same short time > window that we have reconstructed (e.g. 350-400msec) or rather on the whole > epoch (-200-1200 msec) ? > > Thank you very much for your help, > Happy new year! > > Charline > > > * Department of Diagnostic Imaging – Research* > The Hospital for Sick Children (SickKids) > 555 University Ave.Toronto, M5G 1X8 > > Tel: 416-813-7654-309118 > *Email*: [log in to unmask] >