Dear everyone,
> 1. After T1 segmentation , I found the background noise level of bias corrected T1 image was elevated (I attached the corresponded figure of
> original T1 and bias corrected T1 image in this email). Is it normal or something wrong during the bias correction procedure ? In my past
> experience, the bias corrected T1 image looks normal in VBM8 toolbox.
I can't see any attachment, but it seems I observed the same issue with CAT12 r911. During CAT12 preprocessing, an ms file is created (float32, called "Bias corrected", range similar to that of the original structure, see upper image in the attached file), which looks alright / like one would expect for a bias-corrected image. After the "Dartel registration" step, probably during the "Write result map" step, this ms file is updated (now unsigned int16, called "bias and noise corrected, intensity normalized [...]"), and like the description already implies, with a range between 0 and 1 (actually the max is lower, but anyway). This can replicated across subjects acquired with this sequence, for other data I didn't observe any issues like that. If this is of any relevance, the original structure is signed int16. For CAT12 preprocessing I used the defaults except for the resolution options, which were set to native and 1 mm for the normalised files, and the WMHC, which was set to 0 and the corresponding threshold also set to 0.
Best
Helmut
|