Abstract
Groups working with MIMIC still have to rebuild the same link between chest radiographs and the patient context held in the electronic health record. We release MIMIC-CXR-DB, a relational database that joins MIMIC-CXR-JPG radiographs to MIMIC-IV at the level of the hospital admission. Under a common key it exposes a curated cardiopulmonary laboratory panel, ICD-10 primary and secondary diagnoses, and an ontology-grounded triple representation of both radiology and discharge notes against twelve biomedical vocabularies. Existing MIMIC derivatives annotate free text against a fixed list of target findings. Our grounding fixes no such list in advance, which yields 14,863 distinct discharge concepts but costs precision that consumers have to filter. Of 377,095 radiographs (227,827 studies), 146,333 (38.8%) fall within an inpatient admission window in MIMIC-IV v3.1. We keep the rest with a null hadm_id and compare the two subsets, since our observations suggest the linkage rule enriches for admitted, sicker patients. We describe how the database was built, what the physicians reviewed, and what an automated audit shows about the coverage and the concept specificity of the released triples. The database and the build code fall under the existing PhysioNet credentialed access of the parent projects.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MultiTab_015.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=mWS8iyLg7s
BibTex
@InProceedings{TurHou_MIMICCXRDB_MICCAISAT2026,
author = { Turki, Houcemeddine AND Ismaila, Lukman E. AND Ben Salem, Ahmed AND Nebli, Ahmed AND Kchaou, Mahdi AND Dere, Abdulhameed Abiola AND Alzahrani, Anas AND Koubaa, Makram},
title = { { MIMIC-CXR-DB: An Admission-Linked, Ontology-Grounded Multimodal Resource Built on MIMIC-CXR and MIMIC-IV } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17263},
month = {pending},
page = {pending}
}