Skip to content

v3.2.0

Latest
Latest

Choose a tag to compare

View all tags
@kei-lab1106 kei-lab1106 released this 13 Jun 13:59
v3.2.0
80907b6

Overview

OpenMAP-T1 v3.2.0 adds flexible NIfTI output geometry, batch-quality logging for large studies, clearer Docker/GPU documentation, and updated dependencies. README content is expanded and aligned across English, Japanese, and Chinese.

Highlights

Configurable output geometry (--output-space)

Choose how parcellation and intermediate NIfTI files are saved:

  • native (default) — resample to the N4-corrected canonical input grid
  • conform — keep the 1 mm isotropic processing grid (filenames include _1mm)
  • both — write both versions (1 mm outputs under conform/)

CSV regional volumes are always computed on the 1 mm grid, regardless of this option.

Batch QC logging (failed_cases.csv)

At the end of a batch run, OpenMAP-T1 writes failed_cases.csv under OUTPUT_FOLDER when any case:

  • failed during processing
  • was skipped via --only-face-cropping or --only-skull-stripping
  • produced an abnormally small total labeled brain volume (< 10,000 mm³ by default)

Columns: case_id, input_path, status, reason, total_brain_volume_mm3.

Robustness

  • Exit immediately if the input directory does not exist
  • Exit immediately if pretrained model loading fails

Docker & documentation

  • Added .dockerignore for leaner image builds
  • Documented GPU usage in Docker (--gpus all)
  • Expanded FAQ, folder layout, and GPU sections in EN / JA / ZH READMEs
  • Added cover image and Related Research entries

Dependencies

Updated pyproject.toml, uv.lock, requirements.txt, and requirements_for_docker.txt (Python 3.11+, current scientific stack and PyTorch 2.12).

Internal

Reorganized src/utils/ into models/, pipeline/, and output/ for clearer maintenance.

Usage examples

# Default: save NIfTI outputs in native input geometry
uv run python src/parcellation.py -i INPUT -o OUTPUT -m MODEL_FOLDER

# Also save 1 mm grid outputs
uv run python src/parcellation.py -i INPUT -o OUTPUT -m MODEL_FOLDER --output-space both

# Docker with GPU (Linux + NVIDIA)
docker run --rm -it --gpus all -v "$(pwd):/app" openmap-t1 -i input -o output -m model
Assets 2
Loading