Cluster Dice: A simple and fast approach for instance-based semantic segmentation evaluation via many-to-many matching

TY - CHAP

T1 - Cluster Dice

T2 - Medical Imaging 2025: Computer-Aided Diagnosis

AU - Kundu, Soumya Snigdha

AU - Kujawa, Aaron

AU - Ivory, Marina

AU - Barfoot, Theodore

AU - Shapey, Jonathan

AU - Vercauteren, Tom

N1 - Publisher Copyright: © 2025 SPIE.

PY - 2025/4/4

Y1 - 2025/4/4

N2 - In some medical image segmentation tasks, it is important to detect and segment several regions per class, e.g. different lesions of different sizes. This problem is similar to instance segmentation but is typically addressed as a semantic segmentation one followed by a post-processing step to separate the regions. It is key to develop segmentation performance metrics that account for these scenarios, which we coin as instance-sensitive segmentation, where appropriate segmentation of all instances matters. The gold standard performance metric for evaluating standard semantic segmentation tasks is the Dice Similarity Coefficient (DSC). Although its design inherently tackles class imbalance, the DSC fails to faithfully capture segmentation performance across several instances, presenting with instance imbalance. Large instances indeed contribute to the majority of the DSC computation and lead to an inflated performance evaluation. Lesion-wise DSC (Lw-DSC) and the recently proposed CC-Metrics (CC-DSC) inspired from defining predictions instances based on voronoi regions of ground truth instances are approaches that addressed this issue by calculating an individual DSC score for each ground truth instance. These metrics rely on specific segment matching choices to pair predicted and ground-truth instances. In this work, we demonstrate that no single metric can holistically evaluate instance imbalance. Further, we propose Cluster-Dice (C-DSC), one of the first segment matching schemes which allows the merging of ground truth instances. This simple framework is easy to define and implement, transparent, and highly efficient to allow for rapid evaluation, and can accommodate boundary-based metrics as well.

AB - In some medical image segmentation tasks, it is important to detect and segment several regions per class, e.g. different lesions of different sizes. This problem is similar to instance segmentation but is typically addressed as a semantic segmentation one followed by a post-processing step to separate the regions. It is key to develop segmentation performance metrics that account for these scenarios, which we coin as instance-sensitive segmentation, where appropriate segmentation of all instances matters. The gold standard performance metric for evaluating standard semantic segmentation tasks is the Dice Similarity Coefficient (DSC). Although its design inherently tackles class imbalance, the DSC fails to faithfully capture segmentation performance across several instances, presenting with instance imbalance. Large instances indeed contribute to the majority of the DSC computation and lead to an inflated performance evaluation. Lesion-wise DSC (Lw-DSC) and the recently proposed CC-Metrics (CC-DSC) inspired from defining predictions instances based on voronoi regions of ground truth instances are approaches that addressed this issue by calculating an individual DSC score for each ground truth instance. These metrics rely on specific segment matching choices to pair predicted and ground-truth instances. In this work, we demonstrate that no single metric can holistically evaluate instance imbalance. Further, we propose Cluster-Dice (C-DSC), one of the first segment matching schemes which allows the merging of ground truth instances. This simple framework is easy to define and implement, transparent, and highly efficient to allow for rapid evaluation, and can accommodate boundary-based metrics as well.

KW - Dice Similarity

KW - Evaluation Metrics

KW - Instance Imbalance

KW - Segment Matching

KW - Semantic Segmentation

UR - http://www.scopus.com/inward/record.url?scp=105004415657&partnerID=8YFLogxK

U2 - 10.1117/12.3047296

DO - 10.1117/12.3047296

M3 - Conference paper

AN - SCOPUS:105004415657

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2025

A2 - Astley, Susan M.

A2 - Wismuller, Axel

PB - SPIE

Y2 - 17 February 2025 through 20 February 2025

ER -