The continuous Dice coefficient extends the classical binary Dice overlap to probabilistic maps, is bounded by 1 with equality only on perfect overlap, decreases monotonically with mismatch, and shows less size bias and greater robustness to partial-volume effects than the discrete Dice in brain-MRI
Statistical validation of image segmentation quality based on a spatial overlap index.,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Continuous Dice Coefficient: a Method for Evaluating Probabilistic Segmentations
The continuous Dice coefficient extends the classical binary Dice overlap to probabilistic maps, is bounded by 1 with equality only on perfect overlap, decreases monotonically with mismatch, and shows less size bias and greater robustness to partial-volume effects than the discrete Dice in brain-MRI