Dear all !

I know, this has been discussed a lot in the forum, however, I am still not sure, whether my analysis pipeline is correct.

I simply want to compare volumes of FS vs. manual segmentation in the hippocampus and calculate the dice coefficient. So these are my steps:

1. mri_convert T1.mgz T1.nii

2. Register (e.g. with FLIRT) T1.nii to original_mprage.nii (where the manual segmentation was done), output: registration.matrix

3. mri_extract_label aseg.mgz 17 left_FreeSurfer_HC.nii

mri_extract_label 53 right_FreeSurfer_HC.nii

4. Register left_ & right_FreeSurfer_HC.nii to original_mprage.nii by using registration.matrix

5. mri_compute_overlap left_Freesurfer_ HC left_manual_HC

Now here are my questions:

A) I've read that I should use mri_label2vol instead of mri_convert

Is that correct and how would I process that afterwards ?

B) The volumes form aseg.mgz are different (in my case: bigger) to the aseg.stats because of partial volume correction. Can I use partial volume corrected hippocampal masks ?

C) Where can I find information about the segmentation method of the hippocampus, used in FreeSurfer, especially regarding the HC borders ? Because otherwise it might be difficult to calculate and interpret dice coefficients.

Best regards and thanks for any help,

Bernd