Hi,
When running niftilabelsmasker.fit- on Harvard Oxford 1mm maxprob, 50 threshold- I get (volumns,93) shaped data instead of 96 (lateralized) . How can one easily extract the labels for the extracted regions? (usually I would just ignore the first- the background, but if it’s less than (labels_length-1)- one can’t match the ROIs with the labels.
It there a way to get the mapping between the labels and the returned matrix? (the best practice would maybe be to return a dataframe with the labels as column names…)
Thanks!
Hi, it seems to be because the atlas contains some empty regions (e.g. ‘Superior Temporal Gyrus, posterior division’ contains no voxels). but the masker, once fitted, has a labels_ attribute to tell you to which label corresponds each dimension of the masked data. you can get back the label names with np.asarray(atlas.labels)[masker.labels_]
1 Like
Many many thanks for your great help.
you’re welcome! good luck with your project
1 Like