RadImageNet-VQA: A Large-Scale CT and MRI Dataset for Radiologic Visual Question Answering
We introduce RadImageNet-VQA, a large-scale dataset designed for training and benchmarking radiologic VQA on CT and MRI exams. Built from the CT/MRI subset of RadImageNet and its expert-curated anatomical and pathological annotations, RadImageNet-VQA provides 750K images with 7.5M generated samples, including 750K medical captions for visual-text alignment and 6.75M… See the full description on the dataset page:
https://huggingface.co/datasets/raidium/RadImageNet-VQA.
