Probabilistic Modeling of Multi-rater Medical Image Segmentation for Diversity and Personalization

ICLR 2026 Conference Submission21310 Authors

Keywords: probabilistic modeling

Abstract: Medical image segmentation is inherently influenced by data uncertainty, arising from ambiguous boundaries in medical scans and subjective variations among expert annotators. To address this challenge, previous works formulated the multi-rater medical image segmentation task, where multiple experts provide separate annotations for each image. However, existing models are typically constrained to either generating diverse segmentations that lack expert specificity or producing personalized outputs that merely replicate individual annotators. We propose Probabilistic modeling of multi-rater medical image Segmentation (ProSeg) that simultaneously enables both diversification and personalization. Extensive experiments on both the nasopharyngeal carcinoma dataset (NPC) and the lung nodule dataset (LIDC-IDRI) demonstrate that our ProSeg achieves a new state-of-the-art performance, providing segmentation results that are both diverse and expert-personalized.

Primary Area: applications to computer vision, audio, language, and other modalities

Submission Number: 21310