RadFM_data_csv is a collection of files used for training and testing the RadFM foundation model. The dataset includes a radiology test set with captions and article links, a visual question-answering subset for radiology images, and linked article contents. It was authored by chaoyi-wu and last updated on 2024-06-02.
Use Cases
- Training radiology vision-language models based on the RP3D dataset with captions and article links.
- Evaluating visual question-answering performance on radiology images based on the RP3D VQA subdataset.
- Linking radiology cases to explanatory text content based on the provided article.json file.
Strengths
- Includes multiple linked components for multimodal training, such as images, captions, articles, and Q&A pairs.
- Specifically designed for training and testing a radiology foundation model, indicating a focused application.
Limitations
- Description metadata is limited; actual data quality, scale, and column definitions require manual inspection after download.
- Row count, file formats, and license are unknown, which may limit suitability assessment.
Provenance
- Source
- huggingface
- Collection Method
- Data used for training and testing the RadFM model; specific gathering method is not detailed.
- Time Range
- null
- Freshness
- Last updated 2024-06-02 08:10:52.
- Geography
- null