Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning

By E. Tiu, Ellie Talius, Pujan Patel, C. Langlotz, Andrew Y. Ng, Pranav Rajpurkar. Nature Biomedical Engineering, 2022

524 citations recorded by the source corpus.

In tasks involving the interpretation of medical images, suitably trained machine-learning models often exceed the performance of medical experts. Yet such a high-level of performance typically requires that the models be trained with relevant datasets that have been painstakingly annotated by experts. Here we show that a self-supervised model trained on chest X-ray images that lack explicit annotations performs pathology-classification tasks with accuracies comparable to those of radiologists. On an external validation dataset of chest X-rays, the self-supervised model outperformed a fully supervised model in the detection of three pathologies (out of eight), and the performance generalized to pathologies that were not explicitly annotated for model training, to multiple image-interpretation tasks and to datasets from multiple institutions. A self-supervised model trained on chest X-ray images that lack explicit annotations performs pathology-classification tasks with accuracies comparable to those of radiologists.

Subfields: Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/76120de60a9e59c23a372457a056da3c220c64b6

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Radiology, Nuclear Medicine and Imaging

1. Pranav Rajpurkar
2. Jure Leskovec
3. Dimitris N. Metaxas
4. Shaoting Zhang
5. Aaron Y. Lee

Citations in the corpus

• A foundation model for generalizable disease detection from retinal images (2023, background): “For instance, deep neural networks have matched or surpassed the accuracy of clinical experts in various applications 5 , such as referral recommendations for sight-threatening retinal diseases 6 and pathology detection in chest X-ray images 7 .”
• On the Challenges and Perspectives of Foundation Models for Medical Image Analysis (2023, background): “…Ghesu et al. [2022], 3D images (stacks of 2D CT and MR images) Chen et al. [2019], or a particular medical imaging modality which includes X-ray Tiu et al. [2022], CT Huang et al. [2023b], Wang et al. [2023a], endoscopy Wang et al. [2023b], and pathology Chen et al. [2022, 2023b], Vorontsov et…”
• Swin-UMamba: Mamba-based UNet with ImageNet-based pretraining (2024, unknown): “Moreover, there is a need for the scalability and efficiency of Mamba-based models for real-world deployment [32], particularly in resource-constrained environments, which is commonly found in medical practice.”
• Med-Flamingo: a Multimodal Medical Few-shot Learner (2023, background): “Furthermore, many medical applications involve multimodal information, such as radiology tasks that require the analysis of both X-ray images and radiology reports (Tiu et al., 2022).”