Segmentation

AFNI Message Board
1

I would like to use AFINI for both segmentation and resting state analyses. While I’m not familiar with AFNI, it is clear that it is well suited to resting state analyses. But I don’t know if subcortical segmentation (and hence volume estimation) is possible.

I need to be able to run a two phase analysis with segmentation first followed by resting state analysis. The segmentation is aimed at testing reproducibility of volume loss (atrophy) calculation.

Two same-subject (within-subject) scans were taken at baseline only minutes apart, then 2 additional T1 scans of the same subject were take 24 months later. To assess reproducibility of volume change (PVC), derived from segmentation, say of the hippocampus, we will simply take the difference between each pair of within-subject scans:
PVC1= Volume_basline_a - Volume_24months_a/Volume_baseline_a;
PVC2= Volume_basline_b - Volume_24months_b/Volume_baseline_b;
Then median (PVC1- PVC2); closer to zero means higher reproducibility

Any feedback would be much appreciated.

2

It sounds like this might be a more specialized application. You may need to upload some data. PM me with details, and we can discuss.

3

The objective initially is just to assess automated subcortical segmentation (derived volume is the main measure) by how reproducible it is (how similar volume calculation is) using 2 same-subject temporally distinct (by a few minutes at most) T1 images. For example: segment the hippocamapus in subject_A_T1a, then segment the hippcampus of subject_A_T1b; subtract volumes (derived from segmentation) : subject_A_T1a - subject_A_T1b; closer to zero will mean more reproducible. Percentage volume difference could also be used : subject_A_T1a - subject_A_T1b / mean(subject_A_T1a, subject_A_T1b)

In short, I simply want to know if auto segmentation of structures such as the thalamus and hippocampus is possible. If so, then I can assess reproducibility of the segmentation…

Best, Charlie