The current anat2mni qc images are misleading and show systematic differences between registered brain and template, which make it hard to actually detect which registrations went wrong and which not.
Most probably we should either use a different nilearn plotting method or alternatively use FSL slicer to create the qc images. Then the qc images look like the one from func2anat, which would be good.
The current anat2mni qc images are misleading and show systematic differences between registered brain and template, which make it hard to actually detect which registrations went wrong and which not.
Most probably we should either use a different nilearn plotting method or alternatively use FSL slicer to create the qc images. Then the qc images look like the one from func2anat, which would be good.