CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning
Problem
Detecting pneumonia accurately from chest X-rays, a task normally requiring expert radiologist interpretation.
Why it matters
Accurate automated pneumonia detection could reduce dependence on scarce radiologist expertise and speed diagnosis.
What the AI did
CheXNet, a 121-layer convolutional neural network, was trained on ChestX-ray14 (over 100,000 frontal-view X-rays, 14 diseases) and compared against four practicing academic radiologists on an annotated test set.
Results
Exceeded average radiologist performance on the F1 metric; state-of-the-art results on all 14 ChestX-ray14 diseases.
Impact
Highly cited (1220 citations) demonstration of deep learning matching radiologists on chest X-ray diagnosis.
Record
| Venue | arXiv (Cornell University) |
|---|---|
| Published | 2017 |
| Work type | preprint |
| Authors | Pranav Rajpurkar, Jeremy Irvin, Kaylie Zhu, Brandon Yang, Hershel Mehta, Tony Duan, Daisy Yi Ding, Aarti Bagul et al. |
| Citations | 1,220 (OpenAlex, at import time) |
| References | 23 |
| Model or method | CheXNet, a 121-layer convolutional neural network |
| Organization | not stated in source |
| Primary source | https://doi.org/10.48550/arxiv.1711.05225 |
| Open access | green |
| Keywords | deep learning, chest X-ray, pneumonia, radiology, convolutional neural network |
Importance rating
significant — thousands to millions affected
Automated pneumonia detection exceeding radiologists could expand diagnosis where expertise is scarce; abstract demonstrates capability on 100,000+ images but states no disease burden scale.