MedPix is a database of patient cases integrating images and textual information. The content material is organized by disease location (organ system), pathology category, patient profiles, and by image classification and caption. Additional information at https://medpix.nlm.nih.gov/home

MedPix-VQA Dataset

The MedPix-VQA dataset is a version of the data found at MEDPIX-ClinQA, specifically modified to address an image overlap issue that would result from directl splitting the original dataset. This overlap can lead to a model potentially seeing the same image during both training and validation, potentially leading to bias or data leakage.

  Key Modifications:

We have modified the dataset to ensure no image overlap between the training and validation… See the full description on the dataset page: https://huggingface.co/datasets/mmoukouba/MedPix-VQA.

MedPix Grouped QA Dataset

This dataset is a grouped version of MedPix 2.0, a multimodal biomedical dataset, intended for use in training and fine-tuning Visual Question Answering (VQA) and Multimodal LLMs.

  🧩 Structure

Each entry contains:

image: A unique medical image questions: A list of 10 diagnosis-related questions per image answers: A list of corresponding answers

This version reduces the original 20,500 entries (10 Q&A per image) down to 2,050 unique images, with… See the full description on the dataset page: https://huggingface.co/datasets/mmoukouba/MedPix-Grouped-QA.

MedPix 2.0: A Comprehensive Multimodal Biomedical Dataset for Advanced AI Applications.

Please cite our work as follows if you use MedPix 2.0
```
@misc{siragusa2025medpix20comprehensivemultimodal,
title={MedPix 2.0: A Comprehensive Multimodal Biomedical Data set for Advanced AI Applications with Retrieval Augmented Generation and Knowledge Graphs},
author={Irene Siragusa and Salvatore Contino and Massimo La Ciura and Rosario Alicata and Roberto Pirrone},
year={2025},
eprint={2407.02994},
archivePrefix={arXiv},
primaryClass={cs.DB},
url={https://arxiv.org/abs/2407.02994},
}
```

Below a description of Case_topic.json and Descriptions.json is provided. images folder contains all the images of the dataset, while in splitted_dataset folder, a split of the dataset is provided, please refer to /splitted_dataset/README.md for further informations.

Contains a list of JSON, each of these provide the information of a single clinical case. The structure of each element is reported below:

• U_id the UID code idenifies a clinical case

• TAC list of names of the .png files containing the CT scans (if present). Images are under the image folder.

• MRI list of names of the .png files containing the MR scans (if present). Images are under the image folder.

• Case dictionary with the information of the clinical case. It contains the following information:

  • Title the diagnosis
  • History patient's history
  • Exam
  • Findings
  • Differential Diagnosis
  • Case Diagnosis
  • Diagnosis By

• Topic Dictionary with the general information about the disease. It contains the following information:

  • Title the diagnosis
  • Disease Discussion
  • ACR Code
  • Category

  Descriptions.json

Contains a list of JSON, each of these provide the textual information about a single image, stored in the image folder. The structure of each element is reported below:

• Type Can be CT or MR, identifies teh scanning modality of the image.
• U_id The UID code of the clinical case the image belongs to.
• image name of the image file
• location fine-grained information about the body part location of the given image
• location category macro-location of the body-part showen in the given image
• Description Dictionary with the decriptive information of the image. It contains the following information:
  • ACR codes
  • Age age of the patient
  • Sex sex of the patient
  • Caption refers to the specific caption of the image
  • Figure part
  • Modality scanning modality of the image
  • Plane

This dataset tracks the updates made on the dataset "MedPix" as a repository for previous versions of the data and metadata.

dreddyyerram/medpix-clinqa-split dataset hosted on Hugging Face and contributed by the HF Datasets community

Dataset Card for VQA-RAD

  Dataset Description

VQA-RAD is a dataset of question-answer pairs on radiology images. The dataset is intended to be used for training and testing Medical Visual Question Answering (VQA) systems. The dataset includes both open-ended questions and binary "yes/no" questions. The dataset is built from MedPix, which is a free open-access online database of medical images. The question-answer pairs were manually generated by a team of clinicians.… See the full description on the dataset page: https://huggingface.co/datasets/flaviagiammarino/vqa-rad.