Usage
The AutoHS preprocessing workflow takes as principal input the path of the dataset that is to be processed. The input dataset is required to be in valid BIDS format with at least one T1-weighted anatomical series. We highly recommend that you validate your dataset with the free, online BIDS Validator.
The command follows the BIDS Apps convention, similar to QSIPrep and fMRIPrep.
Example
autohs data/bids_root/ out/ participant \
--participant-label 001 \
--fs-license-file /path/to/license.txt \
-w out/work/
Command-Line Arguments
Positional Arguments
bids_dirThe root folder of a BIDS-valid dataset (
sub-*folders at the top level).output_dirThe output path for AutoHS derivatives and visual reports.
analysis_levelProcessing stage to run. AutoHS currently supports only
participant.
Options for filtering BIDS queries
--participant-labelOne or more participant labels without the
sub-prefix (e.g.001002). If omitted, all subjects with T1w data are processed.--session-labelRestrict processing to specific sessions (without the
ses-prefix).--skip-bids-validatorSkip
bids-validator(useful on air-gapped clusters).--bids-filter-fileJSON file with PyBIDS filters for the
t1wquery (QSIPrep-style). Example:{ "t1w": { "session": "1", "run": "1" } }
Options to handle performance
-w,--work-dirWorking directory for intermediate segmentation files. Defaults to
<output_dir>/work.--nthreadsCPU threads for FreeSurfer / FastSurfer.
--runtime {docker,apptainer,auto}Container runtime.
autoprefers Docker when the daemon is available, otherwise Apptainer/Singularity.
Workflow configuration
--fastsurferUse FastSurfer (
--seg_only) instead of FreeSurferrecon-allfor step 1.--fs-license-filePath to FreeSurfer
license.txt. Not required with--fastsurfer.--reports-onlySkip segmentation and re-run AI-compute plus derivative publishing from an existing
-wwork directory (requires prior full run for the same subject/session).--not-stop-on-first-crashContinue processing remaining subjects when one scan fails (default: stop immediately).
Methods boilerplate
--md-only-boilerplatePrint Markdown methods text for manuscripts and exit (NiPreps-style). Combine with
--fastsurferwhen FastSurfer was used. Does not requirebids_dirarguments:python run.py --md-only-boilerplate > methods_autohs.md
Participant mode
AutoHS runs independently per subject (and per session when multiple T1w sessions exist).
Parallel runs can share the same output_dir as long as subject/session labels do not
overlap.
Group level
Group-level analysis is not implemented. AutoHS produces participant-level HS classification and asymmetry metrics only.
Preparing data for AutoHS
AutoHS is a BIDS App: data must follow the BIDS specification for anatomical T1w scans.
Required inputs
sub-<label>/anat/sub-<label>_T1w.nii.gzorsub-<label>/ses-<label>/anat/sub-<label>_ses-<label>_T1w.nii.gz
Supported suffix
Only *_T1w.nii and *_T1w.nii.gz files are used. Other contrasts (T2w, FLAIR)
are ignored.
FreeSurfer license
When using the default FreeSurfer backend, provide a valid FreeSurfer license via
--fs-license-file or by placing license.txt in the AutoHS repository root.