DR_MASTER: 98,750 Fundus Images for Diabetic Retinopathy Grading

98,750 fundus images aggregated from the EyePACS, APTOS, and IDRiD datasets. The collection is labeled for 5-class diabetic retinopathy grading and is released under a CC BY 4.0 license. The original author, organization, and last update date are not specified in the provided metadata.

Use Cases

• Train a 5-class diabetic retinopathy classifier based on the labeled fundus images.
• Benchmark model performance on a multi-source dataset based on the aggregation of EyePACS, APTOS, and IDRiD.
• Conduct research on domain adaptation or generalization across different medical imaging sources mentioned in the description.

Strengths

• Contains 98,750 fundus images, providing a substantial volume for model training.
• Aggregates data from three established sources (EyePACS, APTOS, IDRiD), likely increasing diversity.
• Provides standardized 5-class labels for diabetic retinopathy severity grading.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• Data may reflect geographic or source bias inherent to the aggregated datasets.

Provenance

Source

Aggregated from EyePACS, APTOS, and IDRiD datasets.

Collection Method

Likely compiled from multiple public sources for machine learning.