The results of FreeSurfer include a subcortical segmentation and a
segmentation of white matter only. However, there is no segmentation
of (background), cerebrospinal fluid, gray matter, and white matter
only as produced by other tools. Such segmentation is often (as in our
case) required for an evaluation study.
Thus, relabeling the subcortical segmentation (mri/aseg.mgz) should be
an option. Find attached a C++ implementation of such program,
depending only on the nifticlib library.
a) Does this mapping of FreeSurfer labels to the classes BG, CSF, GM,
and WM look correct? Note that it is mainly based on Table 41 on page
71 of the master thesis of Kowkabzadeh found on the web .
b) Is there already such tool in the FreeSurfer package actually? I
have seen mentioning of mri_compute_volume_fractions on this mailing
list but could not figure how to use it. (A more meaningful help
output of the tools would be nice... in particular what [options] are
 K. Kowkabzadeh, Evaluations of Tissue Segmentation of brain MR
Images, Master of Science Thesis in Biomedical Engineering, Department
of Signals and Systems, Division of Biomedical Engineering, Chalmers
University of Technology, Göteborg, Sweden, 2010 Report No.