bone-contouring 0.3.0 and later is distributed under GPL-3.0-only.
Copyright (c) 2026 Matthias Walle, for original bone-contouring code.
Prior MIT grants remain valid for previously distributed versions; the original
MIT notice is preserved in LICENSES/MIT-previous.txt.

The optional U-Net architecture, preprocessing and morphology are adapted from
Bonelab/HR-pQCT-Segmentation (GPL-3.0), written by Nathan J. Neeteson.
Scientific method and published weights: Nathan J. Neeteson, Bryce A. Besler,
Danielle E. Whittier and Steven K. Boyd, Bone Imaging Laboratory,
University of Calgary.

Neeteson NJ, Besler BA, Whittier DE, Boyd SK (2023). Automatic segmentation of
trabecular and cortical compartments in HR-pQCT images using an embedding-
predicting U-Net and morphological post-processing. Scientific Reports 13, 252.
https://doi.org/10.1038/s41598-022-27350-0
Code: https://github.com/Bonelab/HR-pQCT-Segmentation
Weights: https://zenodo.org/records/14755838

Runtime adaptation: aimio-py I/O, fixed-default inference, device selection,
verified weight cache, NumPy ABI compatibility and transactional publication.
No model retraining or additional contour smoothing/peel is introduced.
