Medical Image De-Identification Benchmark Challenge
Abstract
The de-identification (deID) of protected health information (PHI) and personally identifiable information (PII) is a fundamental requirement for sharing medical images, particularly through public repositories, to ensure compliance with patient privacy laws. In addition, preservation of non-PHI metadata to inform and enable downstream development of imaging artificial intelligence (AI) is an important consideration in biomedical research. The goal of MIDI-B was to provide a standardized platform for benchmarking of DICOM image deID tools based on a set of rules conformant to the HIPAA Safe Harbor regulation, the DICOM Attribute Confidentiality Profiles, and best practices in preservation of research-critical metadata, as defined by The Cancer Imaging Archive (TCIA). The challenge employed a large, diverse, multi-center, and multi-modality set of real de-identified radiology images with synthetic PHI/PII inserted. The MIDI-B Challenge consisted of three phases: training, validation, and test. Eighty individuals registered for the challenge. In the training phase, we encouraged participants to tune their algorithms using their in-house or public data. The validation and test phases utilized the DICOM images containing synthetic identifiers (of 216 and 322 subjects, respectively). Ten teams successfully completed the test phase of the challenge. To measure success of a rule-based approach to image deID, scores were computed as the percentage of correct actions from the total number of required actions. The scores ranged from 97.91% to 99.93%. Participants employed a variety of open-source and proprietary tools with customized configurations, large language models, and optical character recognition (OCR). In this paper we provide a comprehensive report on the MIDI-B Challenge's design, implementation, results, and lessons learned.
Keywords
Cite
@article{arxiv.2507.23608,
title = {Medical Image De-Identification Benchmark Challenge},
author = {Linmin Pei and Granger Sutton and Michael Rutherford and Ulrike Wagner and Tracy Nolan and Kirk Smith and Phillip Farmer and Peter Gu and Ambar Rana and Kailing Chen and Thomas Ferleman and Brian Park and Ye Wu and Jordan Kojouharov and Gargi Singh and Jon Lemon and Tyler Willis and Milos Vukadinovic and Grant Duffy and Bryan He and David Ouyang and Marco Pereanez and Daniel Samber and Derek A. Smith and Christopher Cannistraci and Zahi Fayad and David S. Mendelson and Michele Bufano and Elmar Kotter and Hamideh Haghiri and Rajesh Baidya and Stefan Dvoretskii and Klaus H. Maier-Hein and Marco Nolden and Christopher Ablett and Silvia Siggillino and Sandeep Kaushik and Hongzhu Jiang and Sihan Xie and Zhiyu Wan and Alex Michie and Simon J Doran and Angeline Aurelia Waly and Felix A. Nathaniel Liang and Humam Arshad Mustagfirin and Michelle Grace Felicia and Kuo Po Chih and Rahul Krish and Ghulam Rasool and Nidhal Bouaynaya and Nikolas Koutsoubis and Kyle Naddeo and Kartik Pandit and Tony O'Sullivan and Raj Krish and Qinyan Pan and Scott Gustafson and Benjamin Kopchick and Laura Opsahl-Ong and Andrea Olvera-Morales and Jonathan Pinney and Kathryn Johnson and Theresa Do and Juergen Klenk and Maria Diaz and Arti Singh and Rong Chai and David A. Clunie and Fred Prior and Keyvan Farahani},
journal= {arXiv preprint arXiv:2507.23608},
year = {2025}
}
Comments
19 pages