I have a dataset with 1.5T T1 images which have already been skullstriped.

So to avoid the autorecon1 skullstrip, I ran 'recon-all -i subject.nii.gz -s subject -all -noskullstrip', but it finished with the same error for more than a dozen of images (example log attached). It says that the 'mri_em_register' could not open mask volume brainmask.mgz, which I believe is generated by the skullstrip and indeed could not be found in the 'subject/mri' folder.

How should I ran the pipeline skipping the skullstrip? 

Thank you very much in advance.

Pedro Rosa.