{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,29]],"date-time":"2026-08-29T17:26:29Z","timestamp":1788024389002,"version":"build-2784847793"},"reference-count":30,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,10,7]],"date-time":"2025-10-07T00:00:00Z","timestamp":1759795200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Oncol."],"abstract":"<jats:sec><jats:title>Introduction<\/jats:title><jats:p>Data sharing is essential for advancing research in radiation oncology, particularly for training artificial intelligence (AI) models in medical imaging. However, privacy concerns necessitate de-identification of medical images, including defacing operations to remove facial features. This study evaluates the impact of defacing on AI-driven organ segmentation in head-and-neck (HN) computed tomography (CT) images.<\/jats:p><\/jats:sec><jats:sec><jats:title>Methods<\/jats:title><jats:p>Two defacing algorithms, DeIdentifier and <jats:italic>mri_reface<\/jats:italic>_0.3.3, were applied to 50 patient CT scans. Segmentation accuracy was assessed using two commercially available AI segmentation tools, INTContour and AccuContour<jats:sup>\u00ae<\/jats:sup>, and evaluated using Dice similarity coefficient (DSC), Hausdorff Distance at the 95th percentile (HD95), and Surface Dice Similarity Coefficients (SDSC) with 2 mm tolerance. Dose differences (D0.01cc) were calculated for each structure to evaluate potential clinical implications. Statistical comparisons were made using paired t-tests (p&amp;lt;0.05).<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>The results showed that defacing significantly impacted segmentation of on-face structures (e.g., oral cavity, eyes, lacrimal glands) with reduced DSC (&amp;lt;0.9) and higher HD95 (&amp;gt;2.5 mm), while off-face structures (e.g., brainstem, spinal cord) remained largely unaffected (DSC &amp;gt;0.9, HD95 &amp;lt;2 mm). DeIdentifier better preserved Hounsfield Units (HU) and anatomical consistency than <jats:italic>mri_reface<\/jats:italic>, which introduced more variability, including HU shifts in air regions. Minor differences in segmentation accuracy were observed between defacing algorithms, with mri_reface showing slightly greater variability. AccuContour showed slightly greater segmentation variability than INTContour, particularly for small or complex structures. Dose distribution analysis revealed minimal differences (&amp;lt;20 cGy) in most structures, with the largest variation observed in the Brainstem (34 cGy), followed by Lips_NRG (28 cGy) and Brain (25 cGy).<\/jats:p><\/jats:sec><jats:sec><jats:title>Conclusion<\/jats:title><jats:p>These findings suggest that while defacing alters segmentation accuracy in on-face regions, its overall impact on off-face structures and radiation therapy planning is minimal. Future work should explore domain adaptation techniques to improve model robustness across defaced and non-defaced datasets, ensuring privacy while maintaining segmentation integrity.<\/jats:p><\/jats:sec>","DOI":"10.3389\/fonc.2025.1603593","type":"journal-article","created":{"date-parts":[[2025,10,7]],"date-time":"2025-10-07T04:14:17Z","timestamp":1759810457000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Evaluating the impact of different deface algorithms on deep learning segmentation software performance"],"prefix":"10.3389","volume":"15","author":[{"given":"Ali","family":"Ammar","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Libing","family":"Zhu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"suffix":"IV","given":"Shep","family":"Bryan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nathan Y.","family":"Yu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Carlos","family":"Vargas","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yi","family":"Rong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Quan","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,10,7]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"43","DOI":"10.1016\/j.radonc.2020.09.054","article-title":"From multisource data to clinical decision aids in Radiation Oncology: The need for a clinical data science community","volume":"153","author":"Kazmierska","year":"2020","journal-title":"Radiotherapy Oncol"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.1007\/s10278-013-9622-7","article-title":"The Cancer Imaging Archive (TCIA): Maintaining and operating a public information repository","volume":"26","author":"Clark","year":"2013","journal-title":"J Digital Imaging"},{"key":"B3","doi-asserted-by":"publisher","DOI":"10.2967\/jnmt.119.227819","article-title":"Review of HIPAA, part 1: History, protected health information, and privacy and security rules","volume":"47","author":"Moore","year":"2019","journal-title":"J Nucl Med Technol"},{"key":"B4","doi-asserted-by":"publisher","first-page":"25","DOI":"10.1016\/j.ejmp.2021.02.007","article-title":"Data Preparation for artificial intelligence in Medical Imaging: A comprehensive guide to open-access platforms and Tools","volume":"83","author":"Diaz","year":"2021","journal-title":"Physica Med"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.3390\/healthcare10050755","article-title":"A two-stage de-identification process for privacy-preserving medical image analysis","volume":"10","author":"Shahid","year":"2022","journal-title":"Healthcare"},{"key":"B6","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s10278-024-01182-y","article-title":"Summary of the National Cancer Institute 2023 virtual workshop on medical image de-identification\u2014part 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","volume":"38","author":"Clunie","year":"2024","journal-title":"J Imaging Inf Med"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2017.10.034","article-title":"Image processing and quality control for the first 10,000 brain imaging datasets from UK Biobank","volume":"166","author":"Alfaro-Almagro","year":"2018","journal-title":"NeuroImage"},{"key":"B8","doi-asserted-by":"publisher","first-page":"892","DOI":"10.1002\/hbm.20312","article-title":"A technique for the deidentification of structural brain mr images","volume":"28","author":"Bischoff-Grethe","year":"2007","journal-title":"Hum Brain Mapp"},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.2196\/22739","article-title":"De-identification of facial features in Magnetic Resonance Images: Software Development Using Deep Learning Technology","volume":"22","author":"Jeong","year":"2020","journal-title":"J Med Internet Res"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.1002\/hbm.25639","article-title":"A comparative study between state-of-the-art MRI deidentification and AnonyMI, a new method combining re-identification risk reduction and geometrical preservation","volume":"42","author":"Mikulan","year":"2021","journal-title":"Hum Brain Mapp"},{"key":"B11","doi-asserted-by":"publisher","first-page":"65","DOI":"10.1007\/s12021-012-9160-3","article-title":"Obscuring surface anatomy in Volumetric Imaging Data","volume":"11","author":"Milchenko","year":"2012","journal-title":"Neuroinformatics"},{"key":"B12","article-title":"Quickshear defacing for neuroimages","volume-title":"Proceedings of the 2nd USENIX Conference on Health Security and Privacy, Healthsec\u201911","author":"Schimke","year":"2011"},{"key":"B13","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2021.117845","article-title":"Changing the face of neuroimaging research: Comparing a new MRI de-facing technique with popular alternatives","volume":"231","author":"Schwarz","year":"2021","journal-title":"NeuroImage"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.1016\/j.nicl.2023.103507","article-title":"Effects of de-facing software mri_reface on utility of imaging biomarkers used in alzheimer\u2019s disease research","volume":"40","author":"Schwarz","year":"2023","journal-title":"NeuroImage: Clin"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.3524400","article-title":"Poldracklab\/pydeface: pydeface v2.0.2","author":"Gulban","year":"2022"},{"key":"B16","doi-asserted-by":"publisher","DOI":"10.1002\/alz.093968","article-title":"Comparing de-facing software MiDeFace and mri_reface according to privacy protection and effects on downstream measurements","volume":"20","author":"Schwarz","year":"2024","journal-title":"Alzheimer\u2019s Dementia"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.1117\/1.jmi.10.6.064001","article-title":"Reproducibility evaluation of the effects of MRI defacing on brain segmentation","volume":"10","author":"Gao","year":"2023","journal-title":"J Med Imaging"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.3389\/fradi.2023.1149461","article-title":"Artificial Intelligence in neuroradiology: A scoping review of some ethical challenges","volume":"3","author":"Khosravi","year":"2023","journal-title":"Front Radiol"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.3389\/fonc.2023.1120392","article-title":"Segmentation stability of human head and neck cancer medical images for radiotherapy applications under de-identification conditions: Benchmarking Data Sharing and artificial intelligence use-cases","volume":"13","author":"Sahlsten","year":"2023","journal-title":"Front Oncol"},{"key":"B20","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2023.120289","article-title":"Rood-MRI: Benchmarking the robustness of deep learning segmentation models to out-of-distribution and corrupted data in MRI","volume":"278","author":"Boone","year":"2023","journal-title":"NeuroImage"},{"key":"B21","unstructured":"One pixel attack for fooling deep neural networks\n          \n          \n            \n              Su\n              J\n            \n            \n              Vargas\n              DV\n            \n            \n              Kouichi\n              S\n            \n          \n          \n          2019"},{"key":"B22","doi-asserted-by":"publisher","DOI":"10.2196\/26151","article-title":"Clinically applicable segmentation of head and neck anatomy for radiotherapy: Deep Learning algorithm development and Validation Study","volume":"23","author":"Nikolov","year":"2021","journal-title":"J Med Internet Res"},{"key":"B23","doi-asserted-by":"publisher","DOI":"10.1002\/mp.12197","article-title":"Evaluation of segmentation methods on head and neck CT: Auto-segmentation challenge 2015","volume":"44","author":"Raudaschl","year":"2017","journal-title":"Med Phys"},{"key":"B24","doi-asserted-by":"publisher","DOI":"10.4258\/hir.2020.26.1.61","article-title":"Web-based spine segmentation using deep learning in computed tomography images","volume":"26","author":"Kim","year":"2020","journal-title":"Healthcare Inf Res"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.1016\/j.radonc.2020.05.038","article-title":"Organ at risk delineation for radiation therapy clinical trials: Global Harmonization Group Consensus guidelines","volume":"150","author":"Mir","year":"2020","journal-title":"Radiotherapy Oncol"},{"key":"B26","unstructured":"Tomaroberts\/nii2dcm: Nii2dcm: NIfTI to dicom creation with python"},{"key":"B27","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijrobp.2022.07.914","article-title":"Evaluation of the impact of defacing head-and-neck datasets on Radiomics features and dosimetry","volume":"114","author":"Wang","year":"2022","journal-title":"Int J Radiat OncologyBiologyPhysics"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijrobp.2020.11.011","article-title":"Using auto-segmentation to reduce contouring and dose inconsistency in clinical trials: The simulated impact on RTOG 0617","volume":"109","author":"Thor","year":"2021","journal-title":"Int J Radiat OncologyBiologyPhysics"},{"key":"B29","doi-asserted-by":"publisher","first-page":"024001","DOI":"10.1088\/1361-6560\/ac4000","article-title":"Prospectively-validated deep learning model for segmenting swallowing and chewing structures in CT","volume":"67","author":"Iyer","year":"2022","journal-title":"Phys Med Biol"},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.1177\/15330338221105724","article-title":"Evaluating automatic segmentation for swallowing-related organs for head and neck cancer","volume":"21","author":"Li","year":"2022","journal-title":"Technol Cancer Res Treat"}],"container-title":["Frontiers in Oncology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fonc.2025.1603593\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,7]],"date-time":"2025-10-07T04:14:18Z","timestamp":1759810458000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fonc.2025.1603593\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,7]]},"references-count":30,"alternative-id":["10.3389\/fonc.2025.1603593"],"URL":"https:\/\/doi.org\/10.3389\/fonc.2025.1603593","relation":{},"ISSN":["2234-943X"],"issn-type":[{"value":"2234-943X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,10,7]]},"article-number":"1603593"}}