AutoHS documentation
AutoHS is a BIDS App for automated hippocampal sclerosis (HS) screening from T1-weighted MRI. It is designed for epilepsy surgical workup and research cohorts with BIDS-organized anatomical scans.
What problem does AutoHS solve?
Temporal lobe epilepsy is often associated with hippocampal sclerosis — atrophy and structural change in one hippocampus. Clinicians look for left–right hippocampal volume asymmetry on MRI. AutoHS turns that question into a reproducible, quantitative pipeline:
Segment the T1w scan (FreeSurfer or FastSurfer) to label subcortical structures
Measure left and right hippocampal volumes (mm³)
Compute the asymmetry index (AI): \((L - R) / (L + R)\)
Classify volume laterality and apply published HS screening thresholds
Publish JSON metrics, a clinical summary, and a PDF report as BIDS derivatives
AutoHS is decision-support software, not a standalone diagnosis. Results should be interpreted with clinical context, EEG, and pathology when available.
Start here
New users:
Background and theory — science and theory (clinical background, AI formula, thresholds)
Quick Start — run on sample or your own BIDS dataset in minutes
Installation — Docker, Apptainer, or source install
Researchers writing a paper:
Methods boilerplate — generate methods-section boilerplate (
python run.py --md-only-boilerplate)Citation — required references for AutoHS, IDEAS sample data, and third-party tools