Fmriprep fieldmap estimation

I’m reviewing some output reports following fmriprep and noticed that for some of my participants’ fieldmap estimation, the brainmask outline isn’t quite accurate, especially around the temporal lobes (see below for an example, green line). Does anyone know if this is something to be concerned about or could be impacting the SDC workflow?

It seems like SDC performs ok but might need some improvement. I’m playing around to see if I can improve it myself and will try newer fmriprep versions. I’m not noticing any other errors during processing.

Here is my command (running version 23.1.4 with fmriprep-docker wrapper):
fmriprep-docker $input_dir $output_dir participant --participant-label 002 --me-output-echos --md-only-boilerplate --verbose --fs-subjects-dir $output_dir/sourcedata/freesurfer --fs-license-file /home/optilab/freesurfer/6.0.0/license.txt --output-spaces T1w MNI152NLin6Asym:res-2 func:res-2:res-native fsnative fsaverage fsLR --cifti-output 91k --stop-on-first-crash --mem_mb 80000 --omp-nthreads 8 -w $work"