Hi Jim,
I haven't tried this in a long time, but when I was writing the aseg stuff I implemented something in mri_ca_label to write out the probabilities at each step of the ICM. Run mri_ca_label with "-write_probs <fname>"
and it will write out
sprintf(output_fname, "%s03d.mgz", fname, iteration_number)
is that clear? Not sure it works, but it did at some point :).
Note that in the usual way these are monotonically related to probabilities rather than the actual probabilities themselves.
cheers, Bruce
On Tue, 11 Sep 2007, James N. Porter wrote:
Hello FreeSurfers-
At any point in the recon-all processing stream does a voxel by voxel map of segmentation probability values get generated? If not, are that at all obtainable through post-processing methods?