Evaluation of Vision-Language Models for Detection and Deidentification of Medical Images with Burned-In Protected Health Information

Authors: Taehee Lee, Hyungjin Kim, Seong Ho Park, Seonhye Chae, Soon Ho Yoon

Published: 2025-06-24

DOI: 10.1148/radiol.243664

Source: Full article

---

Abstract

OpenAI’s ChatGPT-4o achieved more than 95% accuracy in detecting burned-in protected health information and obscured it on 81% of medical images, suggesting its potential to improve data privacy and workflow efficiency.