|
Not worry if you haven't done much fMRI analysis before, that means you'll be in an excellent position with a fresh perspective to learn about how to best do these analyses!
My understanding is that it is generally best to do mixed (rather than fixed) effects group analyses (where subjects are treated as a random effects), but I admit to not knowing all the details on when each type of analysis is best in different statistical
scenarios.
The fMRI volume data are all nonlinearly registered to MNI volume standard space. This means that the overall brains will be largely the same size, shape, and position. This does not mean that functionally homologous regions of the cerebral cortex will
be well aligned across subjects. Volume-based registration using T1w images (or, to a lesser extent, surface-based registration with cortical folding patterns) are not very good in achieving these kind of correspondences in many cortical brain areas.
The volume timeseries for each fMRI run is called ${SubjectID}/MNINonLinear/Results/${fMRIName}/${fMRIName}.nii.gz.
As I said above, it's my opinion not to recommend combining across subjects using volume-based registration for the cerebral cortex. There is no guarantee that functionally homologous regions of the cortex will be aligned with volume-based registration.
We are actively working on making better (than folding-based) areal-feature-based surface registrations of the cerebral cortex available, but this will take some time yet before it is publicly available. Before that, we will have grayordinates-based versions
of the task analyses that use FreeSurfer folding-based registration for cerebral cortical surfaces and nonlinear volume registration for subcortical structures. We'll have individual analyses and group mean analyses for two groups available for download
in the next data release (hopefully sometime next week). These group averages will continue to improve with time as we improve the cross-subject alignments on the surface with areal features.
To learn about the new combined surface and volume coordinate system, grayordinates, see this paper:
Peace,
Matt.
From: <Yan>, Shulin <shu.ya...@imperial.ac.uk>
Date: Thursday, September 12, 2013 5:06 PM To: "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: [HCP-Users] multi-subject tfMRI analysis I am beginner of fMRI volume based analysis. I found connectome provides a
minimal preprocessed fMRI volume data which has been normalized and registered to a reference structure (T1W?) for each individual. Activation detection can be directly proceed on this data. My questions are: 1. if I want to do a fixed-effect based multi-subject analysis,
can I just combine the multi-subject data directly? if the fMRI cross subjects are registered to the same reference structure? I found each individual has a T1W data, are they same across all the subjects? 2. which file stores the preprocessed 4D tfMRI data? I found the file names are listed in the document, but no clearly explaination.
Best Shulin _______________________________________________
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ |
- [HCP-Users] multi-subject tfMRI analysis Yan, Shulin
- Re: [HCP-Users] multi-subject tfMRI analysis Glasser, Matthew
- Re: [HCP-Users] multi-subject tfMRI analysis Yan, Shulin
- Re: [HCP-Users] multi-subject tfMRI analysis Glasser, Matthew
