I have a few questions about the May 2014 update to Tracula:
1. Between the 2013-12-04 trac-all version 1.53 and 2014-05-26 trac-all version 1.56, the longitudinal stream now uses the aparc+aseg.mgz from the *.long.* directories, and all output is saved there, too. Are the <subj_tp>.long.<subj_base>/dlabel/anat and anatorig transformed into the base space like other images from the longitudinal recon-all, or are they in the original space of the timepoint T1 image?
2. Are there any changes to the bedpostX step, other than using a different mask? I've already run bedpostx in the Dec 2013 version (with lowb_brain_mask). Can the results be used with the new version by moving the <subj>/dmri.bedpostX directory to <subj_tp>.long.base/dmri.bedpostX?
2a. (Maybe pertains to both versions) The link to .../dmri/nodif_brain_mask.nii.gz seems to be made during the -prep stage using .../dlabel/diff/aparc+aseg_mask.bbr.nii.gz or lowb_brain_mask.nii.gz depending on usemaskanat's value in the dmrirc file. The bedpost stage seems to use the target file as its mask no matter what usemaskanat is in the dmrirc file used when calling trac-all -bedp. (I'm trying to use aparc+aseg_mask for -prep and -path, but lowb_brain_mask for -bedp in anticipation of using the bedpostx output for other seed-base tracking in FSL.)
3. May 2014 version: trac-all -prep -notensor fails when it gets to computing the priors because dmri_train includes the --fa flag and the dtifit_FA.nii.gz is not present. Running trac-all -prior without a computed FA also fails the same way. I could compute the tensor, but thought I would save some time. I didn't realize computing the priors would use it. Does it improve the accuracy?
Thanks, Peggy