{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T01:59:53Z","timestamp":1790301593347,"version":"4.1.0"},"reference-count":34,"publisher":"Springer Science and Business Media LLC","issue":"2","license":[{"start":{"date-parts":[[2025,7,25]],"date-time":"2025-07-25T00:00:00Z","timestamp":1753401600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,7,25]],"date-time":"2025-07-25T00:00:00Z","timestamp":1753401600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["J Digit Imaging. Inform. med."],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>De-identification of medical images is a critical step to ensure privacy during data sharing in research and clinical settings. The initial step in this process involves detecting Protected Health Information (PHI), which can be found in image metadata or imprinted within image pixels. Despite the importance of such systems, there has been limited evaluation of existing AI-based solutions, creating barriers to the development of reliable and robust tools. In this study, we present an AI-based pipeline for PHI detection, comprising three key modules: text detection, text extraction, and text analysis. We benchmark three models\u2014YOLOv11, EasyOCR, and GPT-4o\u2014 across different setups corresponding to these modules, evaluating their performance on two different datasets encompassing multiple imaging modalities and PHI categories. Our findings indicate that the optimal setup involves utilizing dedicated vision and language models for each module, which achieves a commendable balance in performance, latency, and cost associated with the usage of large language models (LLMs). Additionally, we show that the application of LLMs not only involves identifying PHI content but also enhances OCR tasks and facilitates an end-to-end PHI detection pipeline, showcasing promising outcomes through our analysis.<\/jats:p>","DOI":"10.1007\/s10278-025-01619-y","type":"journal-article","created":{"date-parts":[[2025,7,25]],"date-time":"2025-07-25T17:38:38Z","timestamp":1753465118000},"page":"1796-1815","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":3,"title":["Exploring AI-Based System Design for Pixel-Level Protected Health Information Detection in Medical Images"],"prefix":"10.1007","volume":"39","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0284-4269","authenticated-orcid":false,"given":"Tuan","family":"Truong","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ivo M.","family":"Baltruschat","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mark","family":"Klemens","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Grit","family":"Werner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Matthias","family":"Lenga","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2025,7,25]]},"reference":[{"key":"1619_CR1","unstructured":"U.S. Department of Health and Human Services: Guidance regarding methods for de-identification of protected health information in accordance with the health insurance portability and accountability act (HIPAA) privacy rule. Washington DC: U.S. Department of Health and Human Services, 2012."},{"key":"1619_CR2","unstructured":"European Parliament and Council of the European Union: Regulation (EU) 2016\/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95\/46\/EC (General Data Protection Regulation). OJ L 119, 4.5.2016, p. 1\u201388, 2016. Available at:https:\/\/data.europa.eu\/eli\/reg\/2016\/679\/oj. Accessed 13 April 2023."},{"key":"1619_CR3","unstructured":"Health Sector Cybersecurity Coordination Center: A Cost Analysis of Healthcare Sector Data Breaches. US Department of Health and Human Services, Washington, DC, 2019. Available at: https:\/\/www.hhs.gov\/sites\/default\/files\/cost-analysis-of-healthcare-sector-data-breaches.pdf. Accessed 18 June 2025."},{"key":"1619_CR4","unstructured":"Advanced Research Projects Agency for Health (ARPA-H): Innovative Solutions Opening For Imaging Data Exchange (INDEX). Advanced Research Projects Agency for Health (ARPA-H), December 2024. Available at: https:\/\/sam.gov\/opp\/afb05a06fb1043ed99c8f39ab3021cbd\/view. Accessed 9 April 2025."},{"key":"1619_CR5","unstructured":"Kay A: Tesseract: an open-source optical character recognition engine. Linux J 2007(159):2, 2007."},{"key":"1619_CR6","doi-asserted-by":"publisher","unstructured":"Kline A, Appadurai V, Luo Y, Shah S: Medical image deidentification, cleaning and compression using Pylogik. arXiv, https:\/\/doi.org\/10.48550\/arXiv.2304.12322, May 2023. Available at:","DOI":"10.48550\/arXiv.2304.12322"},{"key":"1619_CR7","doi-asserted-by":"publisher","unstructured":"Langlois Q, Szelagowski N, Vanderdonckt J, Jodogne S: Open platform for the de-identification of burned-in texts in medical images using deep learning. In: Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies, Rome, Italy, pp. 297\u2013304. SCITEPRESS - Science and Technology Publications, 2024. https:\/\/doi.org\/10.5220\/0012430300003657.Available at: https:\/\/www.scitepress.org\/DigitalLibrary\/Link.aspx?doi=10.5220\/0012430300003657.","DOI":"10.5220\/0012430300003657."},{"key":"1619_CR8","doi-asserted-by":"publisher","unstructured":"Rempe M, Heine L, Seibold C, H\u00f6rst F, Kleesiek J: De-identification of medical imaging data: A comprehensive tool for ensuring patient privacy. arXiv, https:\/\/doi.org\/10.48550\/arXiv.2410.12402, October 2024. Available at:.","DOI":"10.48550\/arXiv.2410.12402"},{"key":"1619_CR9","doi-asserted-by":"publisher","DOI":"10.1007\/s10278-024-01098-7","author":"JA Macdonald","year":"2024","unstructured":"Macdonald JA, Morgan KR, Konkel B, Abdullah K, Martin M, Ennis C, Lo JY, Stroo M, Snyder DC, Bashir MR: A method for efficient de-identification of DICOM metadata and burned-in pixel text. J Imaging Inform Med, https:\/\/doi.org\/10.1007\/s10278-024-01098-7, 2024.","journal-title":"J Imaging Inform Med"},{"key":"1619_CR10","doi-asserted-by":"publisher","DOI":"10.1007\/s10916-017-0736-1","author":"E Monteiro","year":"2017","unstructured":"Monteiro E, Costa C, Oliveira JL: A de-identification pipeline for ultrasound medical images in DICOM format. J Med Syst, https:\/\/doi.org\/10.1007\/s10916-017-0736-1, 2017.","journal-title":"J Med Syst"},{"key":"1619_CR11","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijmedinf.2019.02.011","author":"P Vcelak","year":"2019","unstructured":"Vcelak P, Kryl M, Kratochvil M, Kleckova J: Identification and classification of DICOM files with burned-in text content. Int J Med Inform, https:\/\/doi.org\/10.1016\/j.ijmedinf.2019.02.011, 2019.","journal-title":"Int J Med Inform"},{"key":"1619_CR12","unstructured":"MD.ai. Available at https:\/\/docs.md.ai\/deid\/deid\/. Accessed 28 April 2025."},{"key":"1619_CR13","unstructured":"Google Cloud. Available at https:\/\/cloud.google.com\/healthcare-api\/docs\/concepts\/de-identification. Accessed 9 December 2024."},{"key":"1619_CR14","unstructured":"John Snow Labs. Available at https:\/\/www.johnsnowlabs.com\/deidentification\/. Accessed 9 December 2024."},{"key":"1619_CR15","unstructured":"Glendor. Available at https:\/\/glendor.com\/. Accessed 24 January 2025."},{"key":"1619_CR16","unstructured":"Microsoft. Presidio, version 2.2.358. Available at https:\/\/microsoft.github.io\/presidio\/. Accessed June 2025."},{"key":"1619_CR17","doi-asserted-by":"publisher","DOI":"10.1016\/j.compbiomed.2024.108189","author":"I Jahan","year":"2024","unstructured":"Jahan I, Laskar MT, Peng C, Huang JX: A comprehensive evaluation of large language models on benchmark biomedical text processing tasks. Comput Biol Med, https:\/\/doi.org\/10.1016\/j.compbiomed.2024.108189, 2024.","journal-title":"Comput Biol Med"},{"key":"1619_CR18","doi-asserted-by":"publisher","unstructured":"Clunie D, Prior F, Rutherford M, Moore S, Parker W, Kondylakis H, Ludwigs C, Klenk J, Lou B, O\u2019Sullivan LT, Marcus D, Dobes J, Gutman A, Farahani K: Summary of the National Cancer Institute 2023 Virtual Workshop on Medical Image De-identification-Part 1: Report of the MIDI Task Group - Best Practices and Recommendations, Tools for Conventional Approaches to De-identification, International Approaches to De-identification, and Industry Panel on Image De-identification. J Imaging Inform Med, https:\/\/doi.org\/10.1007\/s10278-024-01182-y,July 2024.","DOI":"10.1007\/s10278-024-01182-y,"},{"key":"1619_CR19","doi-asserted-by":"publisher","unstructured":"Clunie D, Taylor A, Bisson T, Gutman D, Xiao Y, Schwarz CG, Greve D, Gichoya J, Shih G, Kline A, Kopchick B, Farahani K: Summary of the National Cancer Institute 2023 Virtual Workshop on Medical Image De-identification-Part 2: Pathology Whole Slide Image De-identification, De-facing, the Role of AI in Image De-identification, and the NCI MIDI Datasets and Pipeline. J Imaging Inform Med, https:\/\/doi.org\/10.1007\/s10278-024-01183-x,July 2024.","DOI":"10.1007\/s10278-024-01183-x,"},{"key":"1619_CR20","unstructured":"OpenAI. Available at https:\/\/openai.com\/index\/hello-gpt-4o\/. Accessed 20 December 2024."},{"key":"1619_CR21","doi-asserted-by":"crossref","unstructured":"Wasserthal J, Breit HC, Meyer MT, Pradella M, Hinck D, Sauter AW, Heye T, Boll DT, Cyriac J, Yang S, et al.: TotalSegmentator: robust segmentation of 104 anatomic structures in CT images. Radiol Artif Intell 5(5), 2023.","DOI":"10.1148\/ryai.230024"},{"key":"1619_CR22","doi-asserted-by":"crossref","unstructured":"Huang Z, Pu X, Tang G, Ping M, Jiang G, Wang M, Wei X, Ren Y: BS-80K: The first large open-access dataset of bone scan images. Comput Biol Med 151:106221, 2022.","DOI":"10.1016\/j.compbiomed.2022.106221"},{"key":"1619_CR23","doi-asserted-by":"crossref","unstructured":"Wang X, Peng Y, Lu L, Lu Z, Bagheri M, Summers RM: ChestX-ray8: Hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2097\u20132106, 2017.","DOI":"10.1109\/CVPR.2017.369"},{"key":"1619_CR24","doi-asserted-by":"crossref","unstructured":"Antonelli M, Reinke A, Bakas S, Farahani K, Kopp-Schneider A, Landman BA, Litjens G, Menze B, Ronneberger O, Summers RM, et al.: The medical segmentation decathlon. Nat Commun 13(1):4128, 2022.","DOI":"10.1038\/s41467-022-30695-9"},{"key":"1619_CR25","unstructured":"Farahani K, Clunie D, Klenk J, Kopchick B, Diaz M, Pan Q, Pei L, Prior F, Rutherford M, Singh A, Sutton G, Wagner U: Medical Image De-Identification Benchmark (MIDI-B). Available at https:\/\/www.synapse.org\/Synapse:syn53065760 Accessed 16 April 2025."},{"key":"1619_CR26","unstructured":"MD.ai. Available at https:\/\/www.md.ai. Accessed 28 April 2025."},{"key":"1619_CR27","unstructured":"Ultralytics. Available at https:\/\/docs.ultralytics.com\/models\/yolo11. Accessed 5 December 2024."},{"key":"1619_CR28","doi-asserted-by":"publisher","unstructured":"Redmon J, Divvala S, Girshick R, Farhadi A: You Only Look Once: Unified, Real-Time Object Detection. In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, pp. 779\u2013788, https:\/\/doi.org\/10.1109\/CVPR.2016.91,June 2016.","DOI":"10.1109\/CVPR.2016.91,"},{"key":"1619_CR29","unstructured":"EasyOCR. Available at https:\/\/github.com\/JaidedAI\/EasyOCR. Accessed 20 December 2024."},{"key":"1619_CR30","doi-asserted-by":"publisher","unstructured":"Baek Y, Lee B, Han D, Yun S, Lee H: Character region awareness for text detection. arXiv, https:\/\/doi.org\/10.48550\/arXiv.1904.01941, April 2019.","DOI":"10.48550\/arXiv.1904.01941"},{"key":"1619_CR31","doi-asserted-by":"crossref","unstructured":"Baek J, Kim G, Lee J, Park S, Han D, Yun S, Oh SJ, Lee H: What is wrong with scene text recognition model comparisons? Dataset and model analysis. In: Proceedings of the International Conference on Computer Vision (ICCV), 2019.","DOI":"10.1109\/ICCV.2019.00481"},{"key":"1619_CR32","unstructured":"Brown T, Mann B, Ryder N, Subbiah M, Kaplan JD, Dhariwal P, Neelakantan A, Shyam P, Sastry G, Askell A, et al.: Language models are few-shot learners. Adv Neural Inf Process Syst 33:1877\u20131901, 2020."},{"key":"1619_CR33","unstructured":"Colvin S: Pydantic: Data validation and settings management using Python type annotations, version 2.10.6. Available at https:\/\/github.com\/pydantic\/pydantic. Accessed 28 April 2025."},{"key":"1619_CR34","doi-asserted-by":"publisher","unstructured":"Rutherford MW, Nolan T, Pei L, Wagner U, Pan Q, Farmer P, Smith K, Kopchick B, Opsahl-Ong L, Sutton G, Clunie DA, Farahani K, Prior F: Data in support of the MIDI-B Challenge (MIDI-B-Synthetic-Validation, MIDI-B-Curated-Validation, MIDI-B-Synthetic-Test, MIDI-B-Curated-Test) (Version 1) [Data set]. The Cancer Imaging Archive, https:\/\/doi.org\/10.7937\/cf2p-aw56, 2025","DOI":"10.7937\/cf2p-aw56"}],"container-title":["Journal of Imaging Informatics in Medicine"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10278-025-01619-y.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s10278-025-01619-y","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10278-025-01619-y.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,22]],"date-time":"2026-04-22T16:23:53Z","timestamp":1776875033000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s10278-025-01619-y"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,7,25]]},"references-count":34,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2026,4]]}},"alternative-id":["1619"],"URL":"https:\/\/doi.org\/10.1007\/s10278-025-01619-y","relation":{},"ISSN":["2948-2933"],"issn-type":[{"value":"2948-2933","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,7,25]]},"assertion":[{"value":"29 April 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"10 July 2025","order":2,"name":"revised","label":"Revised","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"10 July 2025","order":3,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"25 July 2025","order":4,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"This research study was conducted retrospectively using human subject data made available in open access. Ethical approval was not required, as confirmed by the license attached to the open-access data.","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Ethics Approval"}},{"value":"The authors declare they were employed by Bayer AG during the time of the research. No other financial or non-financial interests related to the work submitted for publication are disclosed.","order":3,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of Interest"}}]}}