Enhancing Reliability of Medical Image Diagnosis through Top-rank Learning with Rejection Module

2025-08-10

AI Summary

In medical image diagnosis, noisy labels and ambiguous-class samples are prone to being erroneously ranked at the top by conventional top-rank learning, undermining model reliability. To address this, we propose a novel top-rank learning framework integrated with a differentiable rejection module: an auxiliary rejection branch—externally attached to the main network—dynamically quantifies each sample’s deviation from normal patterns, and is jointly optimized with the diagnostic task to enable synergistic training. Crucially, our method requires no additional annotations and autonomously identifies and suppresses anomalous samples’ interference with the ranking objective. Extensive experiments on multiple public medical imaging benchmarks demonstrate substantial improvements in top-k diagnostic accuracy (average +3.2%) and robustness; notably, performance remains stable even under 40% label noise. The framework provides an interpretable, uncertainty-aware filtering mechanism for high-stakes clinical AI systems.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Computer Vision: Medical and Biological Imaging
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

In medical image processing, accurate diagnosis is of paramount importance. Leveraging machine learning techniques, particularly top-rank learning, shows significant promise by focusing on the most crucial instances. However, challenges arise from noisy labels and class-ambiguous instances, which can severely hinder the top-rank objective, as they may be erroneously placed among the top-ranked instances. To address these, we propose a novel approach that enhances toprank learning by integrating a rejection module. Cooptimized with the top-rank loss, this module identifies and mitigates the impact of outliers that hinder training effectiveness. The rejection module functions as an additional branch, assessing instances based on a rejection function that measures their deviation from the norm. Through experimental validation on a medical dataset, our methodology demonstrates its efficacy in detecting and mitigating outliers, improving the reliability and accuracy of medical image diagnoses.

Problem

• Improving medical image diagnosis reliability using top-rank learning
• Addressing noisy labels and class-ambiguous instances in diagnosis
• Mitigating outliers with a rejection module to enhance accuracy

Innovation

• Top-rank learning with rejection module
• Cooptimized top-rank loss and rejection
• Detects and mitigates outliers effectively