SonoVQA is an ultrasound visual question answering (VQA) benchmark with 14,905 expert-annotated QA pairs derived from the UltrasoundCases educational platform. It covers 18 anatomical organs, 6 reasoning levels, and 3 question types across a broad range of ultrasound imaging scenarios.
This is an annotation-only release. The cc-by-nc-4.0 license applies to the SonoVQA annotations, QA pairs, taxonomy labels, splits, and… See the full description on the dataset page:
https://huggingface.co/datasets/QingboKang/SonoVQA.
