Spatial Alignment of Saliency Map with MRI image #7
Description
I have a specific query on the code in get_brainiac_saliency_map.py
Currently, the saliency map generation script saves the output NIfTI files using an identity matrix (np.eye(4)) as the affine header. Could you explain why this is done?
By replacing the original affine matrix with an identity matrix, the spatial metadata (origin, orientation, and voxel spacing) of the original MRI is lost.
In that case, I think the generated saliency maps do not spatially align with the source/raw NIfTI images when loaded in the way done in quickstart.ipynb.