Disease Classification2017en
NIH Clinical Center Chest X-ray Dataset
112,120 frontal-view chest X-ray images from 30,805 unique patients with 14 disease labels extracted using NLP from radiology reports. Foundational benchmark for chest X-ray AI.
Current State of the Art
TorchXRayVision
Cohen Lab
85.8
auroc
auroc Progress Over Time
Showing 3 breakthroughs from May 2017 to Dec 2025
Key Milestones
Nov 2017
CheXNet
Original CheXNet on ChestX-ray14. Exceeded radiologist performance on pneumonia (0.768 vs 0.633).
84.1
+1.8%
Dec 2025
TorchXRayVisionCurrent SOTA
Multi-dataset pre-training improves over single-dataset.
85.8
+2.0%
Total Improvement
3.9%
Time Span
8y 9m
Breakthroughs
3
Current SOTA
85.8
Top Models Performance Comparison
Top 4 models ranked by auroc
Best Score
85.8
Top Model
TorchXRayVision
Models Compared
4
Score Range
5.4
aurocPrimary
| # | Model | Score | Paper / Code | Date |
|---|---|---|---|---|
| 1 | TorchXRayVisionOpen Source Cohen Lab | 85.8 | Dec 2025 | |
| 2 | CheXNetOpen Source Stanford ML Group | 84.1 | Dec 2025 | |
| 3 | DenseNet-121 (Chest X-ray)Open Source Research | 82.6 | Dec 2025 | |
| 4 | ResNet-50 (Chest X-ray)Open Source Research | 80.4 | Dec 2025 |