Dataset · tumor-vqa
Medical Visual Question Answering Benchmark
tumor-vqa/DeepTumorVQA_2.0
3D CT abdominal benchmark for Visual Question Answering with 438K QA pairs and clinical subtypes.
What's inside
Contains 10K to 100K samples (438K total QA pairs, including a 10K curated benchmark and a 428K training pool) across 42 clinical subtypes. Includes pre-extracted 2D and video modalities, as well as 20K agent training trajectories with tool traces in English.
What you can do with it
- Training and evaluating medical multimodal VQA models
- Developing diagnostic AI agents for CT scan analysis
- Tasks
- Visual question answering, Image classification
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 84K
- Likes
- 8
How to load it
from datasets import load_dataset
ds = load_dataset("tumor-vqa/DeepTumorVQA_2.0")