Ranking-Guided Semi-Supervised Domain Adaptation for Severity Classification

2026-04-02

AI Summary

This work addresses the challenge of severity classification in medical images under semi-supervised domain adaptation, where class boundaries are inherently ambiguous yet exhibit a natural ordinal structure. To tackle this, the study proposes a novel framework that explicitly incorporates ordinal class information by jointly aligning cross-domain sample rankings and continuous ranking score distributions. By enforcing consistency in the ordinal semantics of samples across source and target domains, the method effectively mitigates domain shift in ordinal classification tasks. Experimental results on datasets of ulcerative colitis and diabetic retinopathy demonstrate that the proposed approach significantly improves classification accuracy and successfully aligns the ranking distributions across severity levels.

Abstract

Semi-supervised domain adaptation leverages a few labeled and many unlabeled target samples, making it promising for addressing domain shifts in medical image analysis. However, existing methods struggle with severity classification due to unclear class boundaries. Severity classification involves naturally ordered class labels, complicating adaptation. We propose a novel method that aligns source and target domains using rank scores learned via ranking with class order. Specifically, Cross-Domain Ranking ranks sample pairs across domains, while Continuous Distribution Alignment aligns rank score distributions. Experiments on ulcerative colitis and diabetic retinopathy classification validate the effectiveness of our approach, demonstrating successful alignment of class-specific rank score distributions.

Problem

• severity classification
• semi-supervised domain adaptation
• class boundaries
• domain shift
• medical image analysis

Innovation

• Ranking-Guided
• Semi-Supervised Domain Adaptation
• Severity Classification
• Cross-Domain Ranking
• Distribution Alignment