[pymvpa] Include as center_ids only columns (features) which do not contain all 0's

[kade]

Hello everybody!
I'm looking to run a whole-brain searchlight but I want to focus the
"searchlight" on specific voxels - namely ones which are not "empty" space
in my t-maps (the column contains 0 for every sample for a given feature).

I know center_ids is the key, but I'm not sure of the specific way to code
it. The documentation says it requires a "list of feature ids (not
coordinates)", and from looking at other datasets, it looks like that means
it wants a list object (?) with numerical values representing the specific
features I want to use (e.g. a list of [0,1,2] would mean I want to use the
first three features as centers).

The tutorial uses a pre-specified list object (rather than coding one from
scratch), so I'm not quite sure what to do. Am I on the right track, and is
it just basically a list? I learn a lot through examples and it would be
great to get some extra help on this, even though I know it's sort of a
basic question.

Additionally, is there some way to have the searchlight avoid any features
which are "empty" when it grabs a cluster of voxels? I ask because my mask
includes some empty spaces, and it looks like the searchlight includes the
0 columns in the searchlight 'cluster', which affects the overall accuracy
for the center voxel.
Thank you all!
Kade Jentink
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/pkg-exppsy-pymvpa/attachments/20201020/eb3ccdd9/attachment.html>