Hi,
I'm not quite sure what you mean by "trying to combine two separate registrations into a single second-level analysis".
I would also not recommend using an initial_highres for this purpose.
Our standard approach to registration should work fine in this case.
What we recommend is simply that you use the structural from the appropriate scan session in the first level analysis as the "main structural" and then it will register the functional scans to this and then to standard space. All higher-level analyses, including the second level analysis, will be conducted in standard space, although you can always transform your data back into structural space if you want to see it there. Given that all of the final analyses will be run in standard space, this approach is normally fine. If you have tried this and have had problems then please give us some more details about those problems.
All the best,
Mark
On 2 May 2014, at 19:41, Benjamin Philip <[log in to unmask]> wrote:
> I've been trying to complete an analysis where I have 17 subjects, each with 2 sessions (PRE and POST treatment), and 3 runs per session. Based on helpful suggestions in another thread, I'm investigating "PRE-POST" effects by doing a single second-level "subtraction" analysis for each subject, containing all 6 runs (3 from each session), with a "Session" variable set at [-1 -1 -1 1 1 1].
>
> This analysis, however, has highlighted some problems with my registration. We have highres T1 scans from every session, and thus far, I have been registering each functional scan to that session's T1. But the second-level "subtraction" analysis produces some very ugly effects (missing slivers around brain edges), which I suspect are from trying to combine two separate registrations into a single second-level analysis.
>
> I tried using the "expanded structural image" option in FEAT to set things up so the POST sessions would first be registered to the POST structural scan, and then the PRE structural scan would be the "main structural image" for both sessions. But that produces errors like these:
>
> Cannot open volume initial_highres2highres_fast_pve_2 for reading!
> Cannot open volume initial_highres2highres_fast_wmseg for reading!
> Image Exception : #22 :: ERROR: Could not open image initial_highres2highres_fast_wmseg
> ...and many more. Here's a link to my logs/feat5_reg file: https://www.dropbox.com/s/zh1nym160ujrtmo/feat5_reg.zip
>
> What's the best way to register two sets of scans for the same participant, across sessions? I suppose it'd be possible to just ignore the POST structural scan entirely, but that seems sub-optimal if there are any options that can co-register both the PRE and POST brains.
>
> Thanks,
>
> -Benjamin Philip
|