Towards long-tailed, multi-label disease classification from chest X-ray: Overview of the CXR-LT challenge

Abstract

Many real-world image recognition problems, such as diagnostic medical imaging exams, are "long-tailed" $\unicode{x2013}$ there are a few common findings followed by many more relatively rare conditions. In chest radiography, diagnosis is both a long-tailed and multi-label problem, as patients often present with multiple findings simultaneously. While researchers have begun to study the problem of long-tailed learning in medical image recognition, few have studied the interaction of label imbalance and label co-occurrence posed by long-tailed, multi-label disease classification. To engage with the research community on this emerging topic, we conducted an open challenge, CXR-LT, on long-tailed, multi-label thorax disease classification from chest X-rays (CXRs). We publicly release a large-scale benchmark dataset of over 350,000 CXRs, each labeled with at least one of 26 clinical findings following a long-tailed distribution. We synthesize common themes of top-performing solutions, providing practical recommendations for long-tailed, multi-label medical image classification. Finally, we use these insights to propose a path forward involving vision-language foundation models for few- and zero-shot disease classification.

Update after major revision

References in corpus (21)

• YOLOv4: Optimal Speed and Accuracy of Object Detection
• mixup: Beyond Empirical Risk Minimization
• ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases
• Domain-Specific Language Model Pretraining for Biomedical Natural Language Processing
• Ensemble deep learning: A review
• A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises
• PadChest: A large chest x-ray image dataset with multi-label annotated reports
• A Comprehensive Survey of Image Augmentation Techniques for Deep Learning
• Deep Ensembles: A Loss Landscape Perspective
• MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographs
• LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day
• Multi-modal Understanding and Generation for Medical Images and Text via Vision-Language Pre-Training
• Symbolic Discovery of Optimization Algorithms
• RoentGen: Vision-Language Foundation Model for Chest X-ray Generation
• Long-Tailed Classification of Thorax Diseases on Chest X-Ray: A New Benchmark Study
• Battle of the Backbones: A Large-Scale Comparison of Pretrained Models across Computer Vision Tasks
• Simple and Robust Loss Design for Multi-Label Learning with Missing Labels
• Improving Zero-Shot Detection of Low Prevalence Chest Pathologies using Domain Pre-trained Language Models
• Bag of Tricks for Long-Tailed Multi-Label Classification on Chest X-Rays
• How Can We Tame the Long-Tail of Chest X-ray Datasets?
• BB-GCN: A Bi-modal Bridged Graph Convolutional Network for Multi-label Chest X-Ray Recognition

Cited by in corpus (4)

• CLIP in Medical Imaging: A Survey
• Evolution-aware VAriance (EVA) Coreset Selection for Medical Image Classification
• CoRPA: Adversarial Image Generation for Chest X-rays Using Concept Vector Perturbations and Generative Models
• Handling Supervision Scarcity in Chest X-ray Classification: Long-Tailed and Zero-Shot Learning