FL-CheX: Federated Learning Benchmark for Chest X-rays

A benchmark dataset for federated learning research in medical imaging, specifically for chest X-ray analysis. It is structured with 15 simulated client nodes designed to reflect real-world data heterogeneity across hospitals and scanners. The dataset's creator and specific collection dates are not provided in the available metadata.

Use Cases

• Benchmark federated learning algorithms based on the described 15-client simulation.
• Study the impact of demographic heterogeneity on model performance in medical AI.
• Investigate methods to mitigate non-IID data distributions across simulated hospital clients.
• Evaluate model robustness against scanner heterogeneity in medical imaging pipelines.

Strengths

• Designed with 15 client nodes to simulate a federated learning environment.
• Explicitly models three key real-world heterogeneities: demographic, non-IID hospital data, and scanner differences.

Limitations

• Row count and total image volume are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Collection Method

Likely a curated or simulated benchmark for federated learning research.