Federated learning for COVID-19 screening from Chest X-ray images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Public Health

Background:

• The COVID-19 pandemic presents a significant global health challenge.
• Deep learning on chest X-rays shows promise for accelerating COVID-19 diagnosis.
• Medical data privacy regulations hinder centralized data collection for AI model training.

Purpose of the Study:

• To introduce a federated learning framework for collaborative COVID-19 screening from chest X-rays.
• To enable multiple medical institutions to train deep learning models without sharing sensitive patient data.
• To address challenges like non-independent and identically distributed (non-IID) and unbalanced data in federated settings.

Main Methods:

• Developed a collaborative federated learning framework for COVID-19 detection.
• Trained deep learning models across multiple institutions without centralizing patient data.
• Investigated the impact of non-IID and unbalanced data distributions on federated learning performance.

Main Results:

• The proposed federated learning framework achieved competitive performance compared to centralized training.
• Demonstrated the effectiveness of the framework across two different deep learning model architectures.
• Showcased the feasibility of collaborative AI model development under data privacy constraints.

Conclusions:

• Federated learning offers a viable solution for privacy-preserving collaborative AI development in medical imaging.
• The framework facilitates the creation of robust COVID-19 screening models by leveraging distributed datasets.
• Encourages adoption of federated learning to harness the collective power of private medical data for public health.