Dear Jacqueline,
It is difficult to answer this confidently without seeing the images.
However, the edge of the brain may appear to not align well in some cases, but this may not be important, especially in areas where there is signal loss in the EPI. If the internal structures align well, especially the grey-white interface, and you do not have fieldmaps, then I think the result is likely to be the best you will achieve. Make sure you check both the functional to structural and structural to standard registrations separately though. These can highlight problems more clearly.
All the best,
Mark
On 4 Sep 2014, at 20:39, Lutz Jacqueline <[log in to unmask]> wrote:
> Dear experts,
>
> I have integrated the epi-reg command-line registration into FEAT. (The BBR registration in the FEAT-GUI did yield good results, possibly because i don't have fieldmaps). The func2struct yielded better alignment compared to the standard in FEAT, particularly in subcortical areas. So I overwrote example_func2highres.mat and highres2example_func.mat in the reg folder with these bbr results, and run updatefeatreg. But the derived func2standard, look a bit odd:
>
> In some cases, the functional image looks blurred/distorted at the edges and "sticking out" dorsally of the standard brain for 1-2 centimeters. Looking at func2standard with "normal" (not BBR) feat-gui registration, these cases look similarly distorted although alignment did not lead to the sticking out (func always lies within the standardbrain boundaries).
>
> Since the internal structures align much better with the BBR method, I am tempted to go with epi-reg still, but would like your opinion on what might happen in the aforementioned registration approach to result in the different func2standard.
>
> I can send you some pics of what i mean.
>
> Thank you, best
> Jacqueline
|