Hi Sam, > So if I understood correctly, when subjects perform a simple task, then we assume that the underlying neural activity will be similar across subjects. With resting-state, we cannot assume that because subjects will be thinking of different things, is that the rationale? Yep, that's the idea. > I performed the same analysis twice on the same data set using the Melodic gui but the results of these two analyses are not the same. Sometimes, at a given voxel, the voxel values are really very different. Is this to be expected? The order in which the ICs identified by melodic are output is non-deterministic, so you will get a different IC ordering if you run melodic repeatedly on the same data. But you should ultimately get the same set of ICs from different runs - you just need to figure out the corresponding volumes across the different runs. Cheers, Paul On 05/03/2019, Sam W. <[log in to unmask]> wrote: > Thank you for your answer Paul! > So if I understood correctly, when subjects perform a simple task, then we > assume that the underlying neural activity will be similar across subjects. > With resting-state, we cannot assume that because subjects will be thinking > of different things, is that the rationale? > Can I ask you another question please? I performed the same analysis twice > on the same data set using the Melodic gui but the results of these two > analyses are not the same. Sometimes, at a given voxel, the voxel values > are really very different. Is this to be expected? > Sam > > On Mon, Mar 4, 2019 at 11:56 PM paul mccarthy <[log in to unmask]> > wrote: > >> Hi Sam, >> >> > But resting-state sequences normally have a fixed scanning duration, >> > and >> > this is the same for every subject. >> >> We can't assume any sort of similarity in the BOLD time series across >> subjects >> in resting state fMRI, and tensor-ICA does not work well in this >> scenario. So the >> recommended approach is temporal concatenation. >> >> > And what if I want to use fsl_regfilt denoising, which requires the >> single-session >> > ICA approach? Is this not possible with resting-state data? >> >> Yes it is possible - you can perform single-subject ICA-based denoising >> as >> a >> separate preprocessing step for each subject prior to running your group >> ICA. >> >> > But why should we assume that the temporal response is consistent >> > between >> > sessions/subjects in task fmri? Some subjects might have had longer >> scanning >> > times or they might engage the task very differently. >> >> You won't be able to use melodic for group ICA at all if your subjects >> have a different >> number of time points. And I would argue that for most simple >> experimental stimuli, >> you should be able to assume some sort of correspondence of activity >> across subjects >> in the regions of the brain that were involved. Obviously this will >> not hold for more >> complicated stimuli, or stimuli with timings that vary across >> subjects, for which you >> should stick with temporal concatenation. >> >> Cheers, >> >> Paul >> >> On 05/03/2019, Sam W. <[log in to unmask]> wrote: >> > Hi, >> > >> > Can someone please explain to me why we need to select "Multi-session >> > temporal concatenation" for resting-state data? In the Melodic help >> webpage >> > it says that we should use this approach if we "can not assume that the >> > associated temporal response is consistent between sessions/subjects." >> But >> > resting-state sequences normally have a fixed scanning duration, and >> this is >> > the same for every subject. And what if I want to use fsl_regfilt >> denoising, >> > which requires the single-session ICA approach? Is this not possible >> > with >> > resting-state data? >> > For the "Multi-session Tensor-ICA" approach the webpage says "It is >> > recommended to use this approach for data where the stimulus paradigm >> > is >> > consistent between session/subjects.". But why should we assume that >> > the >> > temporal response is consistent between sessions/subjects in task fmri? >> Some >> > subjects might have had longer scanning times or they might engage the >> task >> > very differently. >> > >> > Thank you for your help >> > Sam >> > >> > ######################################################################## >> > >> > To unsubscribe from the FSL list, click the following link: >> > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> > >> >> ######################################################################## >> >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 >> > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1