Medical-Image-Segmentation
Collection
14 items • Updated
2D lung ROI segmentation — SegResNet, sliding-window recipe
Per-slice binary lung segmentation at native resolution. Trained on 256x256 patches, inference via 2D sliding windows (overlap 0.5, Gaussian blending).
Test (45,751 slices): Dice (micro) 0.9823, precision 0.9832, recall 0.9814, per-slice p05 0.8352.
Files: model.pth (weights-only state dict) + config.yaml (full training
configuration). Training code, splits and reports:
wernerp02/nodule-segmentation-demo.
- Downloads last month
- 12
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support