Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Overview of the CXR-LT 2026 Challenge: Multi-Center Long-Tailed and Zero Shot Chest X-ray Classification

Hexin Dong, Yi Lin, Pengyu Zhou, Fengnian Zhao, Alan Clint Legasto, Mingquan Lin, Hao Chen, Yuzhe Yang, George Shih, Yifan Peng

TL;DR

The CXR-LT 2026 challenge introduces a multi-center dataset comprising over 145,000 images from PadChest and NIH Chest X-ray datasets and demonstrates that large-scale vision-language pre-training significantly mitigates the performance drop typically associated with zero-shot diagnosis.

Abstract

Chest X-ray (CXR) interpretation is hindered by the long-tailed distribution of pathologies and the open-world nature of clinical environments. Existing benchmarks often rely on closed-set classes from single institutions, failing to capture the prevalence of rare diseases or the appearance of novel findings. To address this, we present the CXR-LT 2026 challenge. This third iteration of the benchmark introduces a multi-center dataset comprising over 145,000 images from PadChest and NIH Chest X-ray datasets. The challenge defines two core tasks: (1) Robust Multi-Label Classification on 30 known classes and (2) Open-World Generalization to 6 unseen (out-of-distribution) rare disease classes. We report the results of the top-performing teams, evaluating them via mean Average Precision (mAP), AUROC, and F1-score. The winning solutions achieved an mAP of 0.5854 on Task 1 and 0.4315 on Task 2, demonstrating that large-scale vision-language pre-training significantly mitigates the performance drop typically associated with zero-shot diagnosis.

Overview of the CXR-LT 2026 Challenge: Multi-Center Long-Tailed and Zero Shot Chest X-ray Classification

TL;DR

The CXR-LT 2026 challenge introduces a multi-center dataset comprising over 145,000 images from PadChest and NIH Chest X-ray datasets and demonstrates that large-scale vision-language pre-training significantly mitigates the performance drop typically associated with zero-shot diagnosis.

Abstract

Chest X-ray (CXR) interpretation is hindered by the long-tailed distribution of pathologies and the open-world nature of clinical environments. Existing benchmarks often rely on closed-set classes from single institutions, failing to capture the prevalence of rare diseases or the appearance of novel findings. To address this, we present the CXR-LT 2026 challenge. This third iteration of the benchmark introduces a multi-center dataset comprising over 145,000 images from PadChest and NIH Chest X-ray datasets. The challenge defines two core tasks: (1) Robust Multi-Label Classification on 30 known classes and (2) Open-World Generalization to 6 unseen (out-of-distribution) rare disease classes. We report the results of the top-performing teams, evaluating them via mean Average Precision (mAP), AUROC, and F1-score. The winning solutions achieved an mAP of 0.5854 on Task 1 and 0.4315 on Task 2, demonstrating that large-scale vision-language pre-training significantly mitigates the performance drop typically associated with zero-shot diagnosis.
Paper Structure (17 sections, 1 figure, 3 tables)

This paper contains 17 sections, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Dataset Collection
  4. Label Collection and Task Definition
  5. Task 1: Robust Multi-Label Classification
  6. Task 2: Open-World Generalization
  7. Evaluation Metrics
  8. Challenge procedure
  9. Results
  10. Team participation across stages
  11. Task 1: Robust Prediction Results
  12. Task 2: Open-World / OOD Multi-Label Classification
  13. Discussion
  14. The Dominance of Foundation Models
  15. Calibration vs. Discrimination
  16. ...and 2 more sections

Figures (1)

  • Figure 1: Per-class label distribution in the CXR-LT 2026 evaluation sets (development + test).