NiftiSpheresMasker: Account for missing voxels

Hello everyone,

I’m using the NiftiSpheresMasker to calculate a timeseries, which results in data averaged over each sphere.

If I want to check if individual voxels are “missing” (signal is below a threshold for a specific portion of scans) in my data and furthermore maybe even remove the whole parcel, is there a way in which I could achieve that while still using nilearn?

Thanks a lot already!

Do you have a mask of the “missing voxels” somehow ?
Then you may simply averaging mask values per sphere would give you the proportion of missing voxels.

It wouldn’t be a big issue to create one. So you suggest, I’d just use the mask as a single input to the Masker? Seems like a feasible and smart idea, thanks!