COIN: Confidence Score-Guided Distillation for Annotation-Free Cell Segmentation

2025-03-14

AI Summary

Unsupervised cell instance segmentation (UCIS) suffers from inaccurate boundary localization, leading to missed detections and suboptimal performance. To address this, we propose the first fully annotation-free UCIS framework: (1) an unsupervised semantic segmentation module built upon optimal transport to enhance sensitivity to small objects and object boundaries; (2) a prediction-consistency confidence scoring mechanism that leverages multi-view agreement among model predictions as a criterion for pseudo-label quality; and (3) recursive self-distillation to enable high-fidelity instance-level learning. Our framework introduces the novel paradigm of “confidence-guided instance distillation,” achieving precise boundary delineation without any human annotations. Extensive experiments on six benchmark datasets demonstrate consistent superiority over existing unsupervised methods. Notably, it surpasses mainstream semi-supervised and weakly supervised approaches on MoNuSeg and TNBC, with significant improvements in boundary accuracy and recall.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Segmentation
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

Cell instance segmentation (CIS) is crucial for identifying individual cell morphologies in histopathological images, providing valuable insights for biological and medical research. While unsupervised CIS (UCIS) models aim to reduce the heavy reliance on labor-intensive image annotations, they fail to accurately capture cell boundaries, causing missed detections and poor performance. Recognizing the absence of error-free instances as a key limitation, we present COIN (COnfidence score-guided INstance distillation), a novel annotation-free framework with three key steps: (1) Increasing the sensitivity for the presence of error-free instances via unsupervised semantic segmentation with optimal transport, leveraging its ability to discriminate spatially minor instances, (2) Instance-level confidence scoring to measure the consistency between model prediction and refined mask and identify highly confident instances, offering an alternative to ground truth annotations, and (3) Progressive expansion of confidence with recursive self-distillation. Extensive experiments across six datasets show COIN outperforming existing UCIS methods, even surpassing semi- and weakly-supervised approaches across all metrics on the MoNuSeg and TNBC datasets. The code is available at https://github.com/shjo-april/COIN.

Problem

• Reduces reliance on labor-intensive image annotations
• Improves accuracy in capturing cell boundaries
• Identifies error-free instances without ground truth

Innovation

• Unsupervised semantic segmentation with optimal transport
• Instance-level confidence scoring for model prediction
• Progressive expansion via recursive self-distillation