Providing you are not planning a second level analysis that involves making comparisons among groups of subjects, it should be OK to use slightly different approaches to spatially normalise your images - providing you are clear about this in any resulting publications. Using the T1w images for inter-subject alignment will only work well if your fMRI data are not too distorted. If they are distorted, then deformations estimated from the T1w (without large distortions) will be less accurate for your fMRI. For this reason, some people try to segment their fMRI, and use Dartel to align the resulting tissues. Of course, this is unlikely to work for all fMRI scans as they vary considerably in terms of contrast, artifact, resolution, FoV etc. Another option would be to try to segment the T1w scans for those subjects with them, and the T2w scans of the others, and align the GM from the subjects together using Dartel. One slight problem that you may encounter is that the segmentation may identify GM differently in T1w images compared to what it does with T2w images. Best regards, -John On 13 May 2012 21:19, Bob Spunt <[log in to unmask]> wrote: > SPM experts, > > I have an fMRI sample for which the vast majority of subjects have > both a matched bandwidth T2 and high-resolution T1 anatomical volume. > However, a few subjects do not have the T1 volume. I am using the > DARTEL pipeline (new segment, create template, normalise to MNI) to > normalise the functionals. I would like to be able take advantage of > the T1 for the subjects who have it, so I am planning on normalising > the majority of the subjects with the T1, but when doing so I am > unsure how best to deal with the subjects who do not have the T1. Does > it seem acceptable to create a template using the T2 for the entire > sample, but to use this to normalise only the subjects without the T1? > > Thank you in advance for any advice. > > Sincerely, > Bob