 |
 |
Hi, It would be possible to look at voxel-wise changes as long as your registrations are very accurate. Previously you talked about VBM, and I don't think this is what you really mean, as the M stands for morphometry (i.e. analysis of geometry/shape), whereas I think you just mean voxel-wise analysis of diffusion quantities (which is different from VBM). So I will answer your question assuming that you are interested in voxel-wise analysis and not VBM. So the main thing that you need to be very careful about here is to get good registrations, and make sure that you correct for EPI distortions. If you have fieldmaps then it is possible to correct for fieldmaps, but otherwise you will not be able to remove the distortions. This will mean that voxel-wise analysis in the inferior temporal and frontal lobes will be a problem because of the distortions, as they typically vary considerably between subjects. However, uncorrected distortions can also affect the global registration and hence cause problems/biases in all locations. One option that you can try, which will be intermediate between voxel and region analysis, is to smooth the data considerably after registration, which will make it less sensitive to registration errors, although it does not completely remove biases. Note that it will again be most problematic in the inferior frontal and temporal lobes. I hope this helps. All the best, Mark On 27 Jun 2014, at 12:00, Monty Waite <[log in to unmask]> wrote: > Thanks very much Mark > We are looking at DWI images in traumatic brain injury patients - 3 directions, not DTI, ADC maps > Have previously used GM mask from structural applied to DWI space to look for changes in GM ADC between scans. Mean values and then use SPSS to look for statistical significance > Would like to look at this at a voxel level though by registering scans to a standard and then using randomise to assess regional ADC changes between scans > Is this possible? > Thanks > Monty