 |
 |
I'm not sure exactly what procedure would give the best results, but I would suggest realigning all the functional images, specifying the images of different days as different sessions. You can use the "realign only" option for this (because reslicing may not be necessary) - although you could also use the reslice option in order to create a nice mean image. For the anatomical images, you can coregister them all together. Specify the same image in all cases as the reference image. Then coregister one of the anatomical images to one of your realigned fMRI images, selecting all the other anatomical images as "other". How you do spatial normalisation is up to you, but I would do it by segmenting one of the anatomical images (or a mean of all the anatomical images made using ImCalc - expression something like [(i1+i2+i3)./((i1~=0) + (i2~=0) + (i3~=0) + eps)] ). The grey matter can be matched to the grey matter image from the aprior directory, producing an an.mat file that can be applied to all the images for that subject. Best regards, -John > We have a study where we acquired functional data on different days. For > each of the 16 subjects, we have EPI images and T1 images for each of > the days. Within each day the T1 and EPI are aligned, but the EPI 's are > relatively different (full brain on one day, partial on another). Our > question is now, how to preprocess the data in the cleanest way? We > thought of the following (but don't quite know how to implement it): > > step 1: normalise the T1 of the first day -> xxx_sn.mat > step 2: corregister the T1day1 and T1day2 > step 3: apply the same corregistration to all EPIday2 > step 4: normalise the EPIday1 using only the xxx_sn.mat of step 1 and > all the EPIday2 using the same sn.mat but after corregistering them > according to the corregistration function of step 2. > > Does that make sense? How can we combine two functions (normalisation > and corregistration) in one step, without writing the corregistered > images in step 3 and then normalise them - this extra writing probably > decreases image quality. > > Any idea's would be greatly appreciated > > -- > Christian Keysers, PhD > Assistant Professor > > BCN Neuro-Imaging Center > University of Groningen > Antonius Deusinglaan 2 (room 120) > 9713 AW Groningen > > Phone: +31 50 3638794 > Fax: +31 50 3638875