Dear Szabolcs,
On Mon, 4 Sep 2017 18:08:49 +0200, Szabolcs David <[log in to unmask]> wrote:
>Dear CAT experts,
>
>While using CAT realised that for some subjects (typically with bad data
>quality, very aniso voxel size like 1x1x3) the edges of registered atlas
>labels contains other label IDs, like 'rings'. Please see the attached
>images for better understanding:
>CAT_AAL: T1 overlaid with the AAL atlas labels from CAT and thresholded to
>show a particular area, please note the scattered voxels, supposedly those
>are spuriously labeled
>CAT_AAL2: same as above, but showing every area, notice the rims at the
>edges of some labels.
>Happens with every available atlas, not just AAL.
>What could be the issue? In the cat_defaults I changed the
>cat.output.atlas.native = 1, access to expert interface and the affine
>preprocessing to 'full' in the GUI. Please note that for 1x1x1 resolution,
>better data q images the atlas labels are nice and sharp.
Probably you found the answer by yourself. It is difficult to map the atlas information precisely to images with low spatial resolution or large anisotropy due to interpolation effects. However, I will see whether the mapping can be improved for that kind of data.
Best,
Christian
>
>Kind regards,
>Szabolcs
>
|