MedPix-2.0 Dataset Overview

Dataset Introduction

MedPix 2.0 is a comprehensive multimodal biomedical dataset, specifically designed for advanced artificial intelligence applications.

Citation Information

If you use the MedPix 2.0 dataset, please cite it as follows:

@misc{siragusa2024medpix20comprehensivemultimodal, title={MedPix 2.0: A Comprehensive Multimodal Biomedical Dataset for Advanced AI Applications}, author={Irene Siragusa and Salvatore Contino and Massimo La Ciura and Rosario Alicata and Roberto Pirrone}, year={2024}, eprint={2407.02994}, archivePrefix={arXiv}, primaryClass={cs.DB}, url={https://arxiv.org/abs/2407.02994}, }

Dataset Structure

Folder Structure

• images folder: Contains all images in the dataset.
• splitted_dataset folder: Provides a split of the dataset; see /splitted_dataset/README.md for details.

Case_topic.json

Provides a series of JSON objects, each offering information about a clinical case. Each element includes:

• U_id: UID of the clinical case.
• TAC: List of .png file names for CT scans (if any), located in the image folder.
• MRI: List of .png file names for MR scans (if any), located in the image folder.
• Case: Dictionary with clinical case details such as Title, History, Exam, Findings, Differential Diagnosis, Case Diagnosis, Diagnosis By.
• Topic: Dictionary with disease information such as Title, Disease Discussion, ACR Code, Category.

Descriptions.json

Provides a series of JSON objects, each offering textual information for a single image, stored in the image folder. Each element includes:

• Type: CT or MR, indicating the scan modality.
• U_id: UID of the clinical case the image belongs to.
• image: Image file name.
• location: Fine‑grained body part information.
• location category: Macro location of the body part.
• Description: Dictionary with details such as ACR codes, Age, Sex, Caption, Figure part, Modality, Plane.