I would recommend the norm.mgz as it should be scaled so that the WM is 110 and bias fields would have been (mostly) removed.
On 11/12/2024 12:06 AM, Sandhitsu Das wrote:
External Email - Use Caution
We want to perform appearance-based analysis that also takes advantage of automated outputs, especially segmentations. For this, we have considered different options: raw image, orig.mgz, norm.mgz, T1.mgz, in addition to whether to use base or long pipeline images. There are competing considerations of some intensity normalization being helpful, and at the same time avoiding heavy pre-processing. Are there any recommendations from folks who might have done analysis like this ?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer