Using pre-run Freesurfer

[araikes]

I recognize that this concept has been addressed in poldracklab/fmriprep#1182 and I have read that thread prior to posting here. I have previously posted here on this topic but I'll refresh it for the current error. I have the most current version of fmriprep (updated today).

I have run Freesurfer (BIDS/Freesurfer v 6.0) using a longitudinal pipeline for 29 participants. The output from that is currently in OUTDIR/freesurfer. I now am trying to fmriprep. I've run on a single subject in order to assess the usability of the prior freesurfer run. The process completed and I received the reports for the subject. However, none of the freesurfer derivatives are present. A careful inspection of the logs reveals that it finds the pre-run Freesurfer directory but doesn't utilize it.

Here is the error message:

RuntimeError: Command:
recon-all -autorecon1 -noskullstrip -openmp 4 -subjid sub-01 -sd /out/freesurfer -notalairach -nonuintensitycor -nonormalization
Standard output:

ERROR: It appears that this subject ID is an existing
base/template from longitudinal processing (-base):
sub-01
If you are trying to re-run a -base template you
need to pass the -base and all -tp flags:

   \' -base <templateid> -tp <tpNid> ... -all \'

   (Instead of -all you can pass other flags, such
   as -autorecon2 -autorecon3 to run only parts.)

Standard error:

Return code: 1

I am attaching the log file with the error. Here's the commands for fmriprep:

docker run -i --rm \
-v $INDIVDIR:/data:ro \
-v $OUTDIR:/out \
-v /data/BL2/BIDS_derivatives/code/license.txt:/opt/freesurfer/license.txt \
poldracklab/fmriprep \
/data /out \
participant \
--longitudinal \
--low-mem \
--n_cpus 1 \
--output-space template T1w fsaverage \
--work-dir /scratch \
--use-aroma \
--participant_label "sub-$subject" \
--write-graph

Any help would be appreciated

crash_log.txt

[effigies]

I'm sorry, I meant to respond to this earlier. The issue here is going to be that nipype's ReconAll interface evidently doesn't recognize a longitudinal reconstruction as complete, and attempts to resume.

In order to correctly handle this, we'll need to understand more about what the expected output of recon-all with a longitudinal pipeline should be. Could you share the following?

1. The sequence of FreeSurfer commands you use for a single subject, from recon-all -i to complete reconstruction.
2. The full directory structure of a reconstructed longitudinal subject.

If you're feeling ambitious, the place this would need to be changed will be in these two sections:

https://github.com/nipy/nipype/blob/aeb087470481343e4043eca1acf9196538e21d5c/nipype/interfaces/freesurfer/preprocess.py#L942-L1146

https://github.com/nipy/nipype/blob/aeb087470481343e4043eca1acf9196538e21d5c/nipype/interfaces/freesurfer/preprocess.py#L1222-L1290

We'd need ways of identifying a longitudinal subject and a new set of "steps" for the longitudinal analyses.

[pettitta]

Hi, I know I'm late to the game here, but I'm running into the same issue and I can provide you with what you're looking for. Official freesurfer instructions can be found here

For a single participant, with two time points, I first run a cross sectional recon-all for each time point

Time 1:

recon-all -i /BIDS_data/"${SUBID}"/ses-wave1/anat/"${SUBID}"_ses-wave1_T1w.nii.gz -subjid "${SUBID}_ses-wave1" -all

Time 2:

recon-all -i /BIDS_data/"${SUBID}"/ses-wave2/anat/"${SUBID}"_ses-wave2_T1w.nii.gz -subjid "${SUBID}_ses-wave2" -all

Then I created a base template from the cross-sectional scans for the participant:

recon-all -base "${SUBID}" -tp /BIDS_data/derivatives/freesurfer/"${SUBID}"_ses-wave1 -tp /BIDS_data/derivatives/freesurfer/"${SUBID}"_ses-wave2 -all

Finally we use the newly created base to create a longitudinally processed image at each timepoint:

Time 1:

recon-all -long /BIDS_data/derivatives/freesurfer/"${SUBID}"_ses-wave1 "${SUBID}" -all

Time 2:

recon-all -long /BIDS_data/derivatives/freesurfer/"${SUBID}"_ses-wave2 "${SUBID}" -all

I've attached a full directory structure of a longitudinally processed participant. Please let me know if there's anything else you need and thanks for your help!

freesurfer_directory.txt

[araikes]

@pettitta I'm just seeing this right now. I'm curious if this solution works for allowing FMRIPrep/Nipype to identify that Freesurfer has already been run and correctly import the expected folders?

Also, have you tried this with the BIDS/Freesurfer docker, which does all of these steps?

[oesteban]

Was this ever addressed/documented? Thanks for your patience.

[oesteban]

I'll close this and welcome anyone who reopens it because it wasn't correctly addressed.

[effigies]

This isn't done, as it's about using longitudinal FreeSurfer. #21 should cover that.

[oesteban]

Let's then redirect everyone to the longitudinal FS issue you linked.