Running Freesurfer recon-all separately from fmriprep

freesurfer
fmriprep
#1

Hi,

I’m performing Freesurfer’s recon-all prior to running fmriprep, using the following:

    output_dir=$bids_root_dir/derivatives/freesurfer

    #need /bin/perl in order to run the nu_correct portion of recon-all
	export SUBJECTS_DIR=$output_dir
	export FS_LICENSE=${FREESURFER_HOME}/license.txt
	export MINC_BIN_DIR=${FREESURFER_HOME}/mni/bin
	export MINC_LIB_DIR=${FREESURFER_HOME}/mni/lib
	export MY_PERL=$(which perl)

	if [ ! -d $output_dir/sub-${s} ]; then
		#Run recon-all
		if [ -f $bids_root_dir/sub-${s}/anat/sub-${s}*_T2w.nii.gz ]; then
			#Include T2w anatomical image in recon-all processing stream
			recon-all \
			-s sub-{$s} \
			-i $bids_root_dir/sub-${s}/anat/sub-${s}_T1w.nii.gz \
			-T2 $bids_root_dir/sub-${s}/anat/sub-${s}_T2w.nii.gz \
			-T2pial \
			-all \
			-openmp 4
		else
			#No T2w anatomical image in recon-all processing stream
			recon-all \
			-s sub-{$s} \
			-i $bids_root_dir/sub-${s}/anat/sub-${s}_T1w.nii.gz \
			-all \
			-openmp 4
		fi
	fi 

In reading the documentation for surface reconstruction though, I see that there are additional reconstruction steps, inclusion of ANTs brain masks, etc, so I was wondering if running Freesurfer beforehand using my method will produce comparable results compared to fmriprep’s built in method for running Freesurfer?

Thanks

#2

If FreeSurfer does a good job of brain masking, then there will be very little difference to pre-running. If it is a case where either antsBrainExtraction or mri_watershed performs poorly, there will be a noticeable difference. fMRIPrep will register its own T1w template to T1.mgz in either case, to ensure that fMRIPrep interoperates smoothly.

#3

Got it, thanks again!