TorchXRayVision: A library of chest X-ray datasets and models
About
TorchXRayVision is an open source software library for working with chest X-ray datasets and deep learning models. It provides a common interface and common pre-processing chain for a wide set of publicly available chest X-ray datasets. In addition, a number of classification and representation learning models with different architectures, trained on different data combinations, are available through the library to serve as baselines or feature extractors.
Joseph Paul Cohen, Joseph D. Viviano, Paul Bertin, Paul Morrison, Parsa Torabian, Matteo Guarrera, Matthew P Lungren, Akshay Chaudhari, Rupert Brooks, Mohammad Hashir, Hadrien Bertrand• 2021
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Classification | CheXpert (test) | AUC ROC51.5 | 66 | |
| Disease Diagnosis | Open-i | Accuracy78.1 | 41 | |
| Computer-Aided Diagnosis (CAD) | VinDr | AUC0.446 | 32 | |
| Thoracic Disease Classification | ChestX-ray14 | Average Performance50 | 28 | |
| Binary disease diagnosis | VinDr-CXR OOD | Macro Accuracy91.6 | 21 | |
| Binary disease diagnosis | ChestX-ray14 OOD | Macro Acc55.9 | 21 | |
| Binary disease diagnosis | RSNA OOD | Macro Accuracy70.3 | 21 | |
| Binary disease diagnosis | NIH-Google OOD | Macro Accuracy77.4 | 21 | |
| Binary disease diagnosis | CheXpert (test) | Macro Acc68.2 | 21 | |
| Binary disease diagnosis | MIMIC-CXR | Macro Accuracy70.4 | 21 |
Showing 10 of 10 rows