CLoE: Curriculum Learning on Endoscopic Images for Robust MES Classification

2025-08-18

AI Summary

To address label noise (arising from inter-observer variability) and the neglect of ordinal relationships in Mayo Endoscopic Score (MES) classification for ulcerative colitis endoscopic images, this paper proposes CLoE, a curriculum learning framework. Methodologically, CLoE introduces a novel difficulty-aware sample ordering strategy, where image quality—assessed by a lightweight model guided by the Boston Bowel Preparation Scale (BBPS)—serves as a proxy for annotation confidence. It further explicitly models MES’s inherent ordinal structure and integrates ResizeMix augmentation to enhance robustness in both CNN and Transformer architectures. Evaluated on the LIMUC and HyperKvasir datasets, ConvNeXt-Tiny achieves 82.5% accuracy and a weighted kappa of 0.894—outperforming both supervised and self-supervised baselines—while maintaining low computational overhead.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Estimating disease severity from endoscopic images is essential in assessing ulcerative colitis, where the Mayo Endoscopic Subscore (MES) is widely used to grade inflammation. However, MES classification remains challenging due to label noise from inter-observer variability and the ordinal nature of the score, which standard models often ignore. We propose CLoE, a curriculum learning framework that accounts for both label reliability and ordinal structure. Image quality, estimated via a lightweight model trained on Boston Bowel Preparation Scale (BBPS) labels, is used as a proxy for annotation confidence to order samples from easy (clean) to hard (noisy). This curriculum is further combined with ResizeMix augmentation to improve robustness. Experiments on the LIMUC and HyperKvasir datasets, using both CNNs and Transformers, show that CLoE consistently improves performance over strong supervised and self-supervised baselines. For instance, ConvNeXt-Tiny reaches 82.5% accuracy and a QWK of 0.894 on LIMUC with low computational cost. These results highlight the potential of difficulty-aware training strategies for improving ordinal classification under label uncertainty. Code will be released at https://github.com/zeynepozdemir/CLoE.

Problem

• Robust MES classification under label noise
• Addressing ordinal score nature in endoscopic images
• Improving disease severity assessment with curriculum learning

Innovation

• Curriculum learning framework for MES classification
• Image quality as proxy for annotation confidence ordering
• Combines curriculum with ResizeMix augmentation for robustness