Over 75,000 open access and de-identified medical case reports compiled by mauro-nievoff, last updated in January 2024. The dataset includes metadata, clinical case text, image captions, and more than 130,000 images across specialties like oncology, cardiology, surgery, and pathology. Its structure allows mapping images to corresponding article metadata, clinical cases, captions, and labels.
Use Cases
- Train multimodal models for medical image captioning based on the provided captions and images.
- Develop models for clinical case classification based on the text and metadata from specialties like oncology and cardiology.
- Build systems for linking medical images to relevant clinical context based on the dataset's mapping structure.
- Benchmark image-text retrieval algorithms in a medical domain using the paired images and clinical cases.
Strengths
- Contains over 75,000 case reports, providing a substantial corpus.
- Includes more than 130,000 images, offering a large visual component.
- Covers multiple medical specialties, including oncology, cardiology, surgery, and pathology.
- Structure allows mapping images to corresponding metadata, clinical cases, captions, and labels.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Last updated 2024-01-14 15:02:24; freshness should be verified.
Provenance
- Source
- huggingface