QC of fmriprep brain mask and anatomical/temporal CompCor ROIs

Apologies for this novice question, but I’m reviewing output from some fmriprep preprocessing and noticed that for some subjects, the red line demonstrating the brain mask calculated on the BOLD signal is looser then the actual brain (e.g., in the OFC; see attached picture). I understand that some of this may be due to signal loss in the scanner, but is there more to it? How neatly does this contour have to fit the brain? Thanks for any help!

Hi, which fmriprep version are you using?