Hi Clare - Sorry for responding late to any email that was sent while I was away.
The fmajor and slft are sometimes tricky in their initialization b/c of their high curvature. The initial guess for the path is chosen from the atlas and if it's not well-aligned and veers off the subject's white matter somewhere the initialization will fail. In those cases the path.pd.nii.gz will contain only the initial path, so you can overlay it on the subject's FA map and see if that's what's happening.
If you try the reinit trick a few times (if you keep doing it, it'll try different initializations each time) and it doesn't work, you can also try increasing the number of controls points (say, set ncpts = 7) for these cases.
If none of this works, I can try to take a look at one of these data sets.
Hope this helps, a.y
On Fri, 14 Sep 2012, Gibbard, Clare wrote:
Hello,
I have tried out running TRACULA on 20 subjects. It all seemed to work well except that in 8 of the subjects the corpus callosum forceps major was either missing or was a very small cluster. The same happened for the temporal SLF of 2 subjects.
I read previous posts related to this and tried re-running the affected subjects with set reinit =1. However, the results were very similar to before. I have checked the registrations and they are not perfect, but not wildly inaccurate.
Are you able to suggest any options for me to try? I have attached the dmrirc file I used for the first run (called dmrirc-original) as well as the one for re-running the subjects whose CC was not segmented (called dmrirc).
Thank you.
Best wishes, Clare