Dear Helmut,
thank you very much for your answer! Unfortunately we haven't adjusted the FOV for run2 and the images look like that because of the severe movement of the subject as you described in point 1). I guess I have to exclude the data from run2 :-(
I have attached pictures of the raw data, slice time corrected, realigned and mean images of run 1 and run 2 in comparison.( Its very clear that all the pictures of run2 have missing voxels).
But how about subjects where the pictures are not that heavily distorted, but the movement between the sessions is quite severe (1-2cm): should I do the preprocessing together in any case? When would it make sense to do run separate preprocessing batches?
Thanks,
Marta
-----Ursprüngliche Nachricht-----
Von: H. Nebl [mailto:[log in to unmask]]
Gesendet: Freitag, 5. Juli 2013 13:05
An: [log in to unmask]; Czapla, Marta
Betreff: Re: preprocessing of multiple sessions
Dear Marta,
subjects tend to move around a little between runs while "relaxing", but 2 cm is quite much. Leaving this aside, there might be two problems:
1) If you have used the same slice orientations/positions for 1st and 2nd run, or in other words, if you did not correct for these movements by adjusting the FOV, then parts of the brain might be outside of the FOV for the 2nd run. This data is lost in any case. It doesn't help if you preprocess the runs seperately, because at some point you want to compare the 1st with the 2nd run probably, and this model will take into account only voxels above a certain intensity threshold. If there is no data for the 2nd run, then this part is excluded from analysis.
2) In case you have adjusted the FOV for the 2nd run, meaning you coverd the brain in both runs, then problems might have occured during preprocessing due to the misplacement of the 1st relative due to the 2nd run. This can be adjusted by manually reorienting the 2nd run (adjusting for the 2 cm displacement) before you start with preprocessing.
I think 2) is rather improbable as realignment is quite robus in general. But it is difficult to guess, because the mean volume looks normal while in the volume displaying "after realignment", the upper part of the brain is missing. If it's missing in these files (because this part wasn't covered during the 2nd run) it should be missing in the mean volume, and if there were some realignment problems the mean volume should be distorted.
Maybe you can show pictures of the raw data (1st volume of the 1st run and of the 2nd run) as well as the realigned af* files (also 1st volume of the 1st run and of the 2nd run), and once again the mean for comparison?
Best,
Helmut
|