Independent, contamination-aware, statistically-rigorous leaderboard for medical image segmentation models: every score with a bootstrap confidence interval and a ranking-stability test, sliced by failure mode, reproducible with one command. Powered by segauge.
dice benchmark metrics leaderboard evaluation medical-imaging confidence-intervals segmentation reproducibility radiology moose kits amos medical-image-segmentation monai nnunet totalsegmentator hd95 surface-dice ranking-stability
-
Updated
Jun 21, 2026 - Python