 |
 |
Dear Bingjiang, It sounds like you are interested in the reconstruction of power rather than evoked response. In this case, I would suggest that you break down all your trials into shorter segments with the length that would be the common denominator of all your trial lengths (as long as it's not too short). Then you can take these segments as your epochs. In the worst case, you might have to discard some data, it's not a big problem. Best, Vladimir On Mon, Nov 5, 2018 at 5:31 PM Bingjiang Lyu <[log in to unmask]> wrote: > Dear all, > > I'm trying to do MEG source localization with a dataset in which the > length of stimuli varies quite a lot, which ranges from 1.4s to 2.8s. I > wonder if it's possible to specify the time-window of source localization > for each trial separately, so the source localization could be restricted > to the time-window during which the stimuli were presented. It seems that > the time-window is fixed for all trials in the epoched MEG data. > > Or does it make sense to concatenate data of all stimuli to generate one > long trial (which will be 723.6s), and then conduct source localization on > it? > Any advice and suggestions will be greatly appreciated. > > Many thanks. > > Best wishes, > Bingjiang >