Medical-Image-Segmentation
Collection
5 items โข Updated
Per-slice binary lung segmentation at native resolution. Trained on 256x256 patches, inference via 2D sliding windows (overlap 0.5, Gaussian blending).
Test split of the released split.csv (40,505 slices of 113 series): Dice (micro) 0.9805, precision 0.9816, recall 0.9794, per-slice p05 0.8268.
Files: model.pth (weights-only state dict) + config.yaml (full training
configuration). Training and evaluation code, lung bounding boxes and reports:
szegedai/nodule-segmentation-baselines.