CXR-LT 2024: A MICCAI challenge on long-tailed, multi-label, and zero-shot disease classification from chest X-ray

  • 2025-06-09 18:53:31
  • Mingquan Lin, Gregory Holste, Song Wang, Yiliang Zhou, Yishu Wei, Imon Banerjee, Pengyi Chen, Tianjie Dai, Yuexi Du, Nicha C. Dvornek, Yuyan Ge, Zuowei Guo, Shouhei Hanaoka, Dongkyun Kim, Pablo Messina, Yang Lu, Denis Parra, Donghyun Son, Álvaro Soto, Aisha Urooj, René Vidal, Yosuke Yamagishi, Zefan Yang, Ruichi Zhang, Yang Zhou, Leo Anthony Celi, Ronald M. Summers, Zhiyong Lu, Hao Chen, Adam Flanders, George Shih, Zhangyang Wang, Yifan Peng
  • 0

Abstract

The CXR-LT series is a community-driven initiative designed to enhance lungdisease classification using chest X-rays (CXR). It tackles challenges in openlong-tailed lung disease classification and enhances the measurability ofstate-of-the-art techniques. The first event, CXR-LT 2023, aimed to achievethese goals by providing high-quality benchmark CXR data for model developmentand conducting comprehensive evaluations to identify ongoing issues impactinglung disease classification performance. Building on the success of CXR-LT2023, the CXR-LT 2024 expands the dataset to 377,110 chest X-rays (CXRs) and 45disease labels, including 19 new rare disease findings. It also introduces anew focus on zero-shot learning to address limitations identified in theprevious event. Specifically, CXR-LT 2024 features three tasks: (i) long-tailedclassification on a large, noisy test set, (ii) long-tailed classification on amanually annotated "gold standard" subset, and (iii) zero-shot generalizationto five previously unseen disease findings. This paper provides an overview ofCXR-LT 2024, detailing the data curation process and consolidatingstate-of-the-art solutions, including the use of multimodal models for raredisease detection, advanced generative approaches to handle noisy labels, andzero-shot learning strategies for unseen diseases. Additionally, the expandeddataset enhances disease coverage to better represent real-world clinicalsettings, offering a valuable resource for future research. By synthesizing theinsights and innovations of participating teams, we aim to advance thedevelopment of clinically realistic and generalizable diagnostic models forchest radiography.

 

Quick Read (beta)

loading the full paper ...