Hi Darko
what you are asking for is an unsolved problem in brain morphometry. We try very hard to be insensitive/adaptive to MR sequence, but depending on how big your acquisition differences are it is a very, very difficult problem. Do you have the sam esubject(s) scanned on the different protocols? That would at least let you evaulate the differences. And you will certainly want to include site as a nuisance regressor in your final analysis. I expect 6.0 is a bit better than 5.3 in this regard, but it is an empirical question and will depeend on the details of your acquisition.
cheers Bruce
On Tue, 21 Feb 2017, Darko Sarovic wrote:
Hi,
I’m doing volumetric analyses on MR structural data. The problem is that the scanning has been done on different MR machines, as well as with different sequences in some cases. This leads to different absolute values between subjects for volumes and thicknesses when being run through the FreeSurfer segmentation.
My questions regarding this are:
- Is there any way to standardize the results so that the segmentations with different acquisition specs (different MR sequences, MR machine developers) can be comparable?
- I have used v5.3.0 for my segmentations; Is any part of v6.0.0 improved in this regard so that using that version would give more comparable results?
Regards,
Darko Sarovic
Gothenburg University, Sahlgrenska Academy