---
license: cc-by-nc-sa-4.0
---
# CT-RATE-VQA Dataset

We constructed a large-scale **CT-VQA** dataset based on the [ReXGroundingCT](https://huggingface.co/datasets/rajpurkarlab/ReXGroundingCT) data \cite{rexct} to support model training and evaluation.  
For each case, CT volumes were processed along with their corresponding multi-class segmentation masks, where each mask channel represents a specific lesion type.

This dataset is designed for medical visual question answering (Med-VQA) tasks.