Resources


Database Credentialed Access

CXR-PRO: MIMIC-CXR with Prior References Omitted

Vignav Ramesh, Nathan Chi, Pranav Rajpurkar

CXR-PRO is an adaptation of the MIMIC-CXR dataset (consisting of chest radiographs and their associated free-text radiology reports) with references to non-existent priors removed.

generation free-text radiology reports references to priors retrieval large language models

Published: Nov. 23, 2022. Version: 1.0.0

Database Restricted Access

Pulmonary Edema Severity Grades Based on MIMIC-CXR

Ruizhi Liao, Geeticka Chauhan, Polina Golland, et al.

Pulmonary edema metadata and labels for MIMIC-CXR

Published: Feb. 9, 2021. Version: 1.0.1

Database Credentialed Access

MIMIC-CXR-JPG - chest radiographs with structured labels

Alistair Johnson, Matthew Lungren, Yifan Peng, et al.

Chest x-rays in JPG format with structured labels derived from the associated radiology report.

computer vision chest x-ray radiology mimic deep learning

Published: March 12, 2024. Version: 2.1.0

Database Credentialed Access

RadCoref: Fine-tuning coreference resolution for different styles of clinical narratives

Yuxiang Liao, Hantao Liu, Irena Spasic

RadCoref is a small subset of MIMIC-CXR with manually annotated coreference mentions and clusters. Based on the annotated data, we fine-tuned a deep neural model and used it to annotate the whole MIMIC-CXR dataset. Both data are available.

natural language processing coreference resolution radiology

Published: Jan. 30, 2024. Version: 1.0.0

Database Credentialed Access

RadVLM Instruction Dataset

Nicolas Deperrois, Hidetoshi Matsuo, Samuel Ruiperez-Campillo, et al.

This dataset is designed to construct RadVLM, a vision–language model for chest X-ray interpretation. It includes instruction data for tasks such as report generation, abnormality detection, and region grounding, and multitask conversation.

chest x-rays vision-language models medical ai

Published: Sept. 25, 2025. Version: 1.0.0

Database Credentialed Access

MIMIC-Ext-CXR-QBA: A Structured, Tagged, and Localized Visual Question Answering Dataset with Question-Box-Answer Triplets and Scene Graphs for Chest X-ray Images

Philip Müller, Friederike Jungmann, Georgios Kaissis, et al.

We present a large-scale CXR VQA dataset derived from MIMIC-CXR with 42M QA pairs, featuring hierarchical answers, bounding boxes, and structured tags. We generated QA-pairs using LLM-based extraction from radiology reports and localization models.

chest x-rays vqa localization scene graphs

Published: July 22, 2025. Version: 1.0.0

Database Restricted Access

Application of Med-PaLM 2 in the refinement of MIMIC-CXR labels

Kendall Park, Rory Sayres, Andrew Sellergren, et al.

This work further refines the labels associated with CheXpert in MIMIC-CXR-JPG 2.0.0 by filtering with Med-PaLM 2 followed by verification by manual review by three US board-certified radiologists.

mimic-cxr labels

Published: Feb. 4, 2025. Version: 1.0.0

Database Restricted Access

Visual Question Answering evaluation dataset for MIMIC CXR

Timo Kohlberger, Charles Lau, Tom Pollard, et al.

This dataset provides 224 VQAs for 40 test set cases, and 111 VQAs for 23 validation set cases of the MIMIC CXR dataset.

Published: Jan. 28, 2025. Version: 1.0.0

Database Credentialed Access

LLaVA-Rad MIMIC-CXR Annotations

Juan Manuel Zambrano Chaves, Shih-Cheng Huang, Yanbo Xu, et al.

This dataset provides GPT-4 extracted sections of radiology reports from MIMIC-CXR, complementing rule-based section extractions with additional reports with findings, and removing references to priors from findings.

Published: Jan. 24, 2025. Version: 1.0.0

Database Credentialed Access

MS-CXR: Making the Most of Text Semantics to Improve Biomedical Vision-Language Processing

Benedikt Boecking, Naoto Usuyama, Shruthi Bannur, et al.

MS-CXR is a new dataset containing 1162 chest X-ray bounding box labels paired with radiology text descriptions, annotated and verified by two board-certified radiologists.

vision-language processing chest x-ray phrase grounding localization

Published: Nov. 15, 2024. Version: 1.1.0