{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T11:28:51Z","timestamp":1793273331540,"version":"4.5.0"},"reference-count":47,"publisher":"IOP Publishing","issue":"10","license":[{"start":{"date-parts":[[2025,5,16]],"date-time":"2025-05-16T00:00:00Z","timestamp":1747353600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"},{"start":{"date-parts":[[2025,5,16]],"date-time":"2025-05-16T00:00:00Z","timestamp":1747353600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/501100005972","name":"Deutsche Krebshilfe","doi-asserted-by":"crossref","award":["70115909"],"award-info":[{"award-number":["70115909"]}],"id":[{"id":"10.13039\/501100005972","id-type":"DOI","asserted-by":"crossref"}]},{"name":"Munich Medical & Clinician Scientist Program"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Phys. Med. Biol."],"published-print":{"date-parts":[[2025,5,18]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>\n                  <jats:italic>Objective.<\/jats:italic> Uncertainty assessment of deep learning autosegmentation (DLAS) models can support contour corrections in adaptive radiotherapy (ART), e.g. by utilizing Monte Carlo Dropout (MCD) uncertainty maps. However, poorly calibrated uncertainties at the patient level often render these clinically nonviable. We evaluated population-based and patient-specific DLAS accuracy and uncertainty calibration and propose a patient-specific post-training uncertainty calibration method for DLAS in ART. <jats:italic>Approach.<\/jats:italic> The study included 122 lung cancer patients treated with a low-field MR-linac (80\/19\/23 training\/validation\/test cases). Ten single-label 3D-U-Net population-based baseline models (BM) were trained with dropout using planning MRIs (pMRIs) and contours for nine organs-at-riks (OARs) and gross tumor volumes (GTVs). Patient-specific models (PS) were created by fine-tuning BMs with each test patient\u2019s pMRI. Model uncertainty was assessed with MCD, averaged into probability maps. Uncertainty calibration was evaluated with reliability diagrams and expected calibration error (ECE). A proposed post-training calibration method rescaled MCD probabilities for fraction images in BM (calBM) and PS (calPS) after fitting reliability diagrams from pMRIs. All models were evaluated on fraction images using Dice similarity coefficient (DSC), 95th percentile Hausdorff distance (HD95) and ECE. Metrics were compared among models for all OARs combined (<jats:italic>n<\/jats:italic>\u2009=\u2009163), and the GTV (<jats:italic>n<\/jats:italic>\u2009=\u200923), using Friedman and posthoc-Nemenyi tests (<jats:italic>\u03b1<\/jats:italic>\u2009=\u20090.05). <jats:italic>Main results.<\/jats:italic> For the OARs, patient-specific fine-tuning significantly (<jats:italic>p<\/jats:italic>\u2009&lt;\u20090.001) increased median DSC from 0.78 (BM) to 0.86 (PS) and reduced HD95 from 14 mm (BM) to 6.0 mm (PS). Uncertainty calibration achieved substantial reductions in ECE, from 0.25 (BM) to 0.091 (calBM) and 0.22 (PS) to 0.11 (calPS) (<jats:italic>p<\/jats:italic>\u2009&lt;\u20090.001), without significantly affecting DSC or HD95 (<jats:italic>p<\/jats:italic>\u2009&gt;\u20090.05). For the GTV, BM performance was poor (DSC = 0.05) but significantly (<jats:italic>p<\/jats:italic>\u2009&lt;\u20090.001) improved with PS training (DSC = 0.75) while uncertainty calibration reduced ECE from 0.22 (PS) to 0.15 (calPS) (<jats:italic>p<\/jats:italic>\u2009=\u20090.45). <jats:italic>Significance.<\/jats:italic> Post-training uncertainty calibration yields geometrically accurate DLAS models with well-calibrated uncertainty estimates, crucial for ART applications.<\/jats:p>","DOI":"10.1088\/1361-6560\/add640","type":"journal-article","created":{"date-parts":[[2025,5,8]],"date-time":"2025-05-08T22:55:06Z","timestamp":1746744906000},"page":"105018","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":4,"title":["Patient-specific uncertainty calibration of deep learning-based autosegmentation networks for adaptive MRI-guided lung radiotherapy"],"prefix":"10.1088","volume":"70","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-7085-4066","authenticated-orcid":true,"given":"Moritz","family":"Rabe","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1240-3141","authenticated-orcid":false,"given":"Ettore F","family":"Meliad\u00f2","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8121-358X","authenticated-orcid":false,"given":"Sebastian N","family":"Marschner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1287-7825","authenticated-orcid":false,"given":"Claus","family":"Belka","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8709-7252","authenticated-orcid":false,"given":"Stefanie","family":"Corradini","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5565-6889","authenticated-orcid":false,"given":"Cornelis A T","family":"van den Berg","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1707-4068","authenticated-orcid":true,"given":"Guillaume","family":"Landry","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christopher","family":"Kurz","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"266","published-online":{"date-parts":[[2025,5,16]]},"reference":[{"key":"pmbadd640bib1","doi-asserted-by":"publisher","first-page":"243","DOI":"10.1016\/j.inffus.2021.05.008","article-title":"A review of uncertainty quantification in deep learning: techniques, applications and challenges","volume":"76","author":"Abdar","year":"2021","journal-title":"Inf. Fusion"},{"key":"pmbadd640bib2","doi-asserted-by":"publisher","DOI":"10.1259\/bjr.20190594","article-title":"Online daily adaptive proton therapy","volume":"93","author":"Albertini","year":"2019","journal-title":"Br. J. Radiol."},{"key":"pmbadd640bib3","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2021.102101","article-title":"A deep learning-based framework for segmenting invisible clinical target volumes with estimated uncertainties for post-operative prostate cancer radiotherapy","volume":"72","author":"Balagopal","year":"2021","journal-title":"Med. Image Anal."},{"key":"pmbadd640bib4","doi-asserted-by":"publisher","DOI":"10.1002\/acm2.13479","article-title":"Varian Ethos online adaptive radiotherapy for prostate cancer: early results of contouring accuracy, treatment plan quality and treatment time","volume":"23","author":"Byrne","year":"2022","journal-title":"J. Appl. Clin. Med. Phys."},{"key":"pmbadd640bib5","doi-asserted-by":"publisher","first-page":"4971","DOI":"10.1002\/mp.15793","article-title":"Personalized auto-segmentation for magnetic resonance imaging\u2013guided adaptive radiotherapy of prostate cancer","volume":"49","author":"Chen","year":"2022","journal-title":"Med. Phys."},{"key":"pmbadd640bib6","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s13014-019-1308-y","article-title":"MR-guidance in clinical reality: current treatment challenges and future perspectives","volume":"14","author":"Corradini","year":"2019","journal-title":"Radiat. Oncol."},{"key":"pmbadd640bib7","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/978-3-031-73083-2_1","article-title":"CloverNet\u2014leveraging planning annotations for enhanced procedural MR segmentation: an application to adaptive radiation therapy","volume":"vol 15196","author":"De Benetti","year":"2024"},{"key":"pmbadd640bib8","doi-asserted-by":"publisher","first-page":"ubae015","DOI":"10.1093\/bjrai\/ubae015","article-title":"Clinical adoption of deep learning target auto-segmentation for radiation therapy: challenges, clinical risks and mitigation strategies","volume":"1","author":"De Biase","year":"2024","journal-title":"BJR|Artif. Intell."},{"key":"pmbadd640bib9","doi-asserted-by":"publisher","first-page":"12","DOI":"10.2307\/2987588","article-title":"The comparison and evaluation of forecasters","volume":"32","author":"Degroot","year":"1983","journal-title":"J. R. Stat. Soc. D"},{"key":"pmbadd640bib10","article-title":"Leveraging uncertainty estimates for predicting segmentation quality","author":"DeVries","year":"2018"},{"key":"pmbadd640bib11","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s00066-024-02262-2","article-title":"Deep learning for autosegmentation for radiotherapy treatment planning: state-of-the-art and novel perspectives","volume":"201","author":"Erdur","year":"2024","journal-title":"Strahlenther. Onkol."},{"key":"pmbadd640bib12","article-title":"Bayesian convolutional neural networks with Bernoulli approximate variational inference","author":"Gal","year":"2016a"},{"key":"pmbadd640bib13","first-page":"pp 1050","article-title":"Dropout as a Bayesian approximation: representing model uncertainty in deep learning","author":"Gal","year":"2016b"},{"key":"pmbadd640bib14","doi-asserted-by":"publisher","first-page":"e11","DOI":"10.1016\/j.prro.2020.07.003","article-title":"Time analysis of online adaptive magnetic resonance\u2013guided radiation therapy workflow according to anatomical sites","volume":"11","author":"G\u00fcng\u00f6r","year":"2021","journal-title":"Pract. Radiat. Oncol."},{"key":"pmbadd640bib15","first-page":"pp 1321","article-title":"On calibration of modern neural networks","author":"Guo","year":"2017"},{"key":"pmbadd640bib16","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s13014-020-01571-x","article-title":"MR-guided proton therapy: a review and a preview","volume":"15","author":"Hoffmann","year":"2020","journal-title":"Radiat. Oncol."},{"key":"pmbadd640bib17","doi-asserted-by":"publisher","first-page":"457","DOI":"10.1007\/s10994-021-05946-3","article-title":"Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods","volume":"110","author":"H\u00fcllermeier","year":"2021","journal-title":"Mach. Learn."},{"key":"pmbadd640bib18","doi-asserted-by":"publisher","first-page":"1","DOI":"10.3389\/fnins.2020.00282","article-title":"Analyzing the quality and challenges of uncertainty estimations for brain tumor segmentation","volume":"14","author":"Jungo","year":"2020","journal-title":"Front. Neurosci."},{"key":"pmbadd640bib19","doi-asserted-by":"publisher","first-page":"48","DOI":"10.1007\/978-3-030-32245-8_6","article-title":"Assessing reliability and challenges of uncertainty estimations for medical image segmentation","volume":"vol 11765","author":"Jungo","year":"2019"},{"key":"pmbadd640bib20","doi-asserted-by":"publisher","first-page":"1573","DOI":"10.1002\/mp.16056","article-title":"Patient-specific transfer learning for auto-segmentation in adaptive 0.35 T MRgRT of prostate cancer: a bi-centric evaluation","volume":"50","author":"Kawula","year":"2023","journal-title":"Med. Phys."},{"key":"pmbadd640bib21","doi-asserted-by":"publisher","first-page":"2295","DOI":"10.1002\/mp.17580","article-title":"Personalized deep learning auto-segmentation models for adaptive fractionated magnetic resonance-guided radiation therapy of the abdomen","volume":"52","author":"Kawula","year":"2025","journal-title":"Med. Phys."},{"key":"pmbadd640bib22","doi-asserted-by":"crossref","DOI":"10.5244\/C.31.57","article-title":"Bayesian SegNet: model uncertainty in deep convolutional encoder-decoder architectures for scene understanding","author":"Kendall","year":"2017"},{"key":"pmbadd640bib23","first-page":"pp 1","article-title":"What uncertainties do we need in Bayesian deep learning for computer vision?","author":"Kendall","year":"2017"},{"key":"pmbadd640bib24","doi-asserted-by":"publisher","first-page":"371","DOI":"10.1007\/978-3-030-12029-0_40","article-title":"Left-ventricle quantification using residual U-Net","volume":"vol 11395","author":"Kerfoot","year":"2019"},{"key":"pmbadd640bib25","doi-asserted-by":"publisher","first-page":"98","DOI":"10.1016\/j.ctro.2019.04.007","article-title":"Technical design and concept of a 0.35 T MR-Linac","volume":"18","author":"Kl\u00fcter","year":"2019","journal-title":"Clin. Transl. Radiat. Oncol."},{"key":"pmbadd640bib26","doi-asserted-by":"publisher","DOI":"10.1016\/j.phro.2024.100535","article-title":"A review of the clinical introduction of 4D particle therapy research concepts","volume":"29","author":"Kn\u00e4usl","year":"2024","journal-title":"Phys. Imaging Radiat. Oncol."},{"key":"pmbadd640bib27","first-page":"pp 1","article-title":"Simple and scalable predictive uncertainty estimation using deep ensembles","volume":"vol 30","author":"Lakshminarayanan","year":"2017"},{"key":"pmbadd640bib28","doi-asserted-by":"publisher","first-page":"222","DOI":"10.1016\/j.radonc.2022.11.004","article-title":"Patient-specific daily updated deep learning auto-segmentation for MRI-guided adaptive radiotherapy","volume":"177","author":"Li","year":"2022","journal-title":"Radiother. Oncol."},{"key":"pmbadd640bib29","doi-asserted-by":"publisher","first-page":"82","DOI":"10.1002\/acm2.13399","article-title":"Prospects for daily online adaptive radiotherapy via Ethos for prostate cancer patients without nodal involvement using unedited CBCT auto-segmentation","volume":"22","author":"Moazzezi","year":"2021","journal-title":"J. Appl. Clin. Med. Phys."},{"key":"pmbadd640bib30","doi-asserted-by":"publisher","first-page":"2901","DOI":"10.1609\/aaai.v29i1.9602","article-title":"Obtaining well calibrated probabilities using Bayesian binning","volume":"vol 29","author":"Naeini","year":"2015"},{"key":"pmbadd640bib31","first-page":"pp 38","article-title":"Measuring calibration in deep learning","volume":"vol 2","author":"Nixon","year":"2019"},{"key":"pmbadd640bib32","doi-asserted-by":"publisher","first-page":"22TR01","DOI":"10.1088\/1361-6560\/ac344f","article-title":"Adaptive proton therapy","volume":"66","author":"Paganetti","year":"2021","journal-title":"Phys. Med. Biol."},{"key":"pmbadd640bib33","doi-asserted-by":"publisher","first-page":"825","DOI":"10.1007\/s11548-022-02597-0","article-title":"Can uncertainty estimation predict segmentation performance in ultrasound bone imaging?","volume":"17","author":"Pandey","year":"2022","journal-title":"Int. J. Comput. Assist. Radiol. Surg."},{"key":"pmbadd640bib34","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s00066-024-02277-9","article-title":"Artificial intelligence for treatment delivery: image-guided radiotherapy","volume":"201","author":"Rabe","year":"2024","journal-title":"Strahlenther. Onkol."},{"key":"pmbadd640bib35","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6560\/ad682d","article-title":"Enhancing the reliability of deep learning-based head and neck tumour segmentation using uncertainty estimation with multi-modal images","volume":"69","author":"Ren","year":"2024","journal-title":"Phys. Med. Biol."},{"key":"pmbadd640bib36","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s13014-023-02330-4","article-title":"Deep learning based automatic segmentation of organs-at-risk for 0.35 T MRgRT of lung tumors","volume":"18","author":"Ribeiro","year":"2023","journal-title":"Radiat. Oncol."},{"key":"pmbadd640bib37","doi-asserted-by":"publisher","DOI":"10.1016\/j.phro.2023.100500","article-title":"A network score-based metric to optimize the quality assurance of automatic radiotherapy target segmentations","volume":"28","author":"Rodr\u00edguez Outeiral","year":"2023","journal-title":"Phys. Imaging Radiat. Oncol."},{"key":"pmbadd640bib38","doi-asserted-by":"publisher","first-page":"234","DOI":"10.1007\/978-3-319-24574-4_28","article-title":"U-Net: convolutional networks for biomedical image segmentation","volume":"vol 9351","author":"Ronneberger","year":"2015"},{"key":"pmbadd640bib39","doi-asserted-by":"publisher","first-page":"1052","DOI":"10.1109\/ISBI48211.2021.9434131","article-title":"Post training uncertainty calibration of deep networks for medical image segmentation","author":"Rousseau","year":"2021"},{"key":"pmbadd640bib40","article-title":"An uncertainty-aware loss function for training neural networks with calibrated predictions","author":"Shamsi","year":"2023"},{"key":"pmbadd640bib41","doi-asserted-by":"publisher","DOI":"10.1016\/j.radonc.2024.110667","article-title":"Incorporating patient-specific prior clinical knowledge to improve clinical target volume auto-segmentation generalisability for online adaptive radiotherapy of rectal cancer: a multicenter validation","volume":"203","author":"Silv\u00e9rio","year":"2025","journal-title":"Radiother. Oncol."},{"key":"pmbadd640bib42","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2020.101693","article-title":"Embracing imperfect datasets: a review of deep learning solutions for medical image segmentation","volume":"63","author":"Tajbakhsh","year":"2020","journal-title":"Med. Image Anal."},{"key":"pmbadd640bib43","doi-asserted-by":"publisher","first-page":"304","DOI":"10.1016\/j.semradonc.2022.06.001","article-title":"Uncertainty assessment for deep learning radiotherapy applications","volume":"32","author":"van den Berg","year":"2022","journal-title":"Semin. Radiat. Oncol."},{"key":"pmbadd640bib44","doi-asserted-by":"publisher","DOI":"10.1016\/j.adro.2021.100658","article-title":"Using spatial probability maps to highlight potential inaccuracies in deep learning-based contours: facilitating online adaptive radiation therapy","volume":"6","author":"van Rooij","year":"2021","journal-title":"Adv. Radiat. Oncol."},{"key":"pmbadd640bib45","doi-asserted-by":"publisher","first-page":"34","DOI":"10.1016\/j.neucom.2019.01.103","article-title":"Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks","volume":"338","author":"Wang","year":"2019","journal-title":"Neurocomputing"},{"key":"pmbadd640bib46","doi-asserted-by":"publisher","first-page":"54","DOI":"10.1016\/j.ctro.2019.04.001","article-title":"Adaptive radiotherapy: the Elekta Unity MR-linac concept","volume":"18","author":"Winkel","year":"2019","journal-title":"Clin. Transl. Radiat. Oncol."},{"key":"pmbadd640bib47","doi-asserted-by":"publisher","first-page":"152","DOI":"10.1016\/j.radonc.2019.10.019","article-title":"Comparing deep learning-based auto-segmentation of organs at risk and clinical target volumes to expert inter-observer variability in radiotherapy planning","volume":"144","author":"Wong","year":"2020","journal-title":"Radiother. Oncol."}],"container-title":["Physics in Medicine &amp; Biology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,5,16]],"date-time":"2025-05-16T10:34:40Z","timestamp":1747391680000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/add640"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,5,16]]},"references-count":47,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2025,5,16]]},"published-print":{"date-parts":[[2025,5,18]]}},"URL":"https:\/\/doi.org\/10.1088\/1361-6560\/add640","relation":{},"ISSN":["0031-9155","1361-6560"],"issn-type":[{"value":"0031-9155","type":"print"},{"value":"1361-6560","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,5,16]]},"assertion":[{"value":"Patient-specific uncertainty calibration of deep learning-based autosegmentation networks for adaptive MRI-guided lung radiotherapy","name":"article_title","label":"Article Title"},{"value":"Physics in Medicine & Biology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2025 The Author(s). Published on behalf of Institute of Physics and Engineering in Medicine by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2025-01-22","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2025-05-08","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2025-05-16","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}