The problem with saving morphed labels to an annot file

✉️ Mailing List Archive (read-only)
1

Dear group,

It seems like an old problem, but I couldn't find a reasonable solution,

When morphing labels from fsaverage to a subject, in most cases the annot file cannot be written due to vertices that belong to more than one label.

I've implemented several solutions, but in some cases, none of them gives a reasonable result.

Thanks,

Noam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20180328/4f5c717d/attachment.html

2

hi Noam,

I personally don't have a solution for this. Maybe someone else?

could you provide a code snippet to replicate the pb?

Alex

3

Hey Alex,

Yes, sure!

You can find the code here:

https://github.com/pelednoam/mmvt/blob/master/src/misc/labels_morphing_problem.py

I'm using the Lausanne125 parcellation (http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0048121), but this problem can also be replicated using the aparc.DKTatlas40 atlas (or almost any other atlas I assume).

If you wish to run it with the Lausanne125 parcellation, the fsaverage annot files can be found here:

https://www.dropbox.com/sh/qvpvzf35konkwe5/AADVCp24TxLa_SBUSAAvCuFMa?dl=0

Thanks!

Noam

4

Hey Noam,

Does this code guarantee that the resulting morphed labels do not overlap?

Eric

5

Hey Eric,

No, I've just used label.morph() for each label from fsaverage to sample subject.

That is actually my question, if someone has a good approach to avoid these overlaps.

I'm using a naive approach (which also depends on the order of the labels) that uses a vertices->labels mapping:

for label in labels:
    for vertice in label.vertices:
        lookup[hemi][vertice] = labels_names.index(label.name)

And then set the labels vertices:

for vertice, label in lookup[hemi].items():
    labels_vertices[label].append(vertice)

I'm wondering if someone has implemented a better solution.

Thanks!

Noam