{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,2]],"date-time":"2026-10-02T16:55:51Z","timestamp":1790960151571,"version":"4.1.0"},"reference-count":49,"publisher":"Wiley","issue":"8","license":[{"start":{"date-parts":[[2022,6,13]],"date-time":"2022-06-13T00:00:00Z","timestamp":1655078400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100006775","name":"GE Healthcare","doi-asserted-by":"publisher","award":["BlueSkyAward"],"award-info":[{"award-number":["BlueSkyAward"]}],"id":[{"id":"10.13039\/100006775","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["aapm.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Medical Physics"],"published-print":{"date-parts":[[2022,8]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Prostate cancer remains the second deadliest cancer for American men despite clinical advancements. Currently, magnetic resonance imaging (MRI) is considered the most sensitive non\u2010invasive imaging modality that enables visualization, detection, and localization of prostate cancer, and is increasingly used to guide targeted biopsies for prostate cancer diagnosis. However, its utility remains limited due to high rates of false positives and false negatives as well as low inter\u2010reader\u00a0agreements.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>Machine learning methods to detect and localize cancer on prostate MRI can help standardize radiologist interpretations. However, existing machine learning methods vary not only in model architecture, but also in the ground truth labeling strategies used for model training. We compare different labeling strategies and the effects they have on the performance of different machine learning models for prostate cancer detection on\u00a0MRI.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Four different deep learning models (SPCNet, U\u2010Net, branched U\u2010Net, and DeepLabv3+) were trained to detect prostate cancer on MRI using 75 patients with radical prostatectomy, and evaluated using 40 patients with radical prostatectomy and 275 patients with targeted biopsy. Each deep learning model was trained with four different label types: pathology\u2010confirmed radiologist labels, pathologist labels on whole\u2010mount histopathology images, and lesion\u2010level and pixel\u2010level digital pathologist labels (previously validated deep learning algorithm on histopathology images to predict pixel\u2010level Gleason patterns) on whole\u2010mount histopathology images. The pathologist and digital pathologist labels (collectively referred to as pathology labels) were mapped onto pre\u2010operative MRI using an automated MRI\u2010histopathology registration\u00a0platform.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Radiologist labels missed cancers (ROC\u2010AUC: 0.75\u20100.84), had lower lesion volumes (~68% of pathology lesions), and lower Dice overlaps (0.24\u20100.28) when compared with pathology labels. Consequently, machine learning models trained with radiologist labels also showed inferior performance compared to models trained with pathology labels. Digital pathologist labels showed high concordance with pathologist labels of cancer (lesion ROC\u2010AUC: 0.97\u20101, lesion Dice: 0.75\u20100.93). Machine learning models trained with digital pathologist labels had the highest lesion detection rates in the radical prostatectomy cohort (aggressive lesion ROC\u2010AUC: 0.91\u20100.94), and had generalizable and comparable performance to pathologist label\u2010trained\u2010models in the targeted biopsy cohort (aggressive lesion ROC\u2010AUC: 0.87\u20100.88), irrespective of the deep learning architecture. Moreover, machine learning models trained with pixel\u2010level digital pathologist labels were able to selectively identify aggressive and indolent cancer components in mixed lesions on MRI, which is not possible with any human\u2010annotated label\u00a0type.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Machine learning models for prostate MRI interpretation that are trained with digital pathologist labels showed higher or comparable performance with pathologist label\u2010trained models in both radical prostatectomy and targeted biopsy cohort. Digital pathologist labels can reduce challenges associated with human annotations, including labor, time, inter\u2010 and intra\u2010reader variability, and can help bridge the gap between prostate radiology and pathology by enabling the training of reliable machine learning models to detect and localize prostate cancer on\u00a0MRI.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1002\/mp.15777","type":"journal-article","created":{"date-parts":[[2022,5,28]],"date-time":"2022-05-28T11:44:36Z","timestamp":1653738276000},"page":"5160-5181","update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":15,"title":["Bridging the gap between prostate radiology and pathology through machine learning"],"prefix":"10.1002","volume":"49","author":[{"given":"Indrani","family":"Bhattacharya","sequence":"first","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"},{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David S.","family":"Lim","sequence":"additional","affiliation":[{"name":"Department of Computer Science Stanford University Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Han Lin","family":"Aung","sequence":"additional","affiliation":[{"name":"Department of Biomedical Data Science Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xingchen","family":"Liu","sequence":"additional","affiliation":[{"name":"Department of Biomedical Data Science Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Arun","family":"Seetharaman","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering Stanford University Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Christian A.","family":"Kunder","sequence":"additional","affiliation":[{"name":"Department of Pathology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wei","family":"Shao","sequence":"additional","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Simon J. C.","family":"Soerensen","sequence":"additional","affiliation":[{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"},{"name":"Department of Epidemiology and Population Health Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Richard E.","family":"Fan","sequence":"additional","affiliation":[{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pejman","family":"Ghanouni","sequence":"additional","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"},{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Katherine J.","family":"To'o","sequence":"additional","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"},{"name":"Department of Radiology VA Palo Alto Health Care System Palo Alto California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"James D.","family":"Brooks","sequence":"additional","affiliation":[{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Geoffrey A.","family":"Sonn","sequence":"additional","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"},{"name":"Department of Urology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mirabela","family":"Rusu","sequence":"additional","affiliation":[{"name":"Department of Radiology Stanford University School of Medicine Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2022,6,13]]},"reference":[{"key":"e_1_2_9_2_1","doi-asserted-by":"publisher","DOI":"10.3322\/caac.21654"},{"key":"e_1_2_9_3_1","doi-asserted-by":"publisher","DOI":"10.1016\/S0140-6736(16)32401-1"},{"key":"e_1_2_9_4_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.cct.2015.02.008"},{"key":"e_1_2_9_5_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eururo.2018.11.031"},{"key":"e_1_2_9_6_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eururo.2018.11.023"},{"key":"e_1_2_9_7_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eururo.2013.03.025"},{"key":"e_1_2_9_8_1","doi-asserted-by":"publisher","DOI":"10.1056\/NEJMoa1801993"},{"key":"e_1_2_9_9_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.urology.2018.04.007"},{"key":"e_1_2_9_10_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.euf.2017.11.010"},{"key":"e_1_2_9_11_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eururo.2015.08.038"},{"key":"e_1_2_9_12_1","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.2020190646"},{"key":"e_1_2_9_13_1","doi-asserted-by":"crossref","unstructured":"StolkTT deJongIJ KweeTC et\u00a0al.False positives in PIRADS (V2) 3 4 and 5 lesions: relationship with reader experience and zonal location.2019;44:1044\u20101051.","DOI":"10.1007\/s00261-019-01919-2"},{"key":"e_1_2_9_14_1","doi-asserted-by":"publisher","DOI":"10.1002\/jmri.23618"},{"key":"e_1_2_9_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2014.2303821"},{"key":"e_1_2_9_16_1","doi-asserted-by":"publisher","DOI":"10.1186\/s12880-019-0308-6"},{"key":"e_1_2_9_17_1","doi-asserted-by":"publisher","DOI":"10.18383\/j.tom.2018.00033"},{"key":"e_1_2_9_18_1","doi-asserted-by":"publisher","DOI":"10.1117\/1.JMI.5.4.044507"},{"key":"e_1_2_9_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2019.2901928"},{"key":"e_1_2_9_20_1","first-page":"552","article-title":"An automated two\u2010step pipeline for aggressive prostate lesion detection from multi\u2010parametric MR sequence","volume":"2020","author":"Sanyal J","year":"2020","journal-title":"AMIA Summits Transl Sci Proc"},{"key":"e_1_2_9_21_1","doi-asserted-by":"crossref","unstructured":"BhattacharyaI SeetharamanA ShaoW et\u00a0al.CorrSigNet: learning correlated prostate cancer signatures from radiology and pathology images for improved computer aided diagnosis. In:International Conference on Medical Image Computing and Computer\u2010Assisted Intervention.Springer;2020:315\u2010325.","DOI":"10.1007\/978-3-030-59713-9_31"},{"key":"e_1_2_9_22_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.14855"},{"key":"e_1_2_9_23_1","unstructured":"SahaA HosseinzadehM HuismanH.End\u2010to\u2010end prostate cancer detection in bpMRI via 3D CNNs: effect of attention mechanisms clinical priori and decoupled false positive reduction. arXiv:2101.03244.2021."},{"key":"e_1_2_9_24_1","doi-asserted-by":"crossref","unstructured":"YuX et\u00a0al.Deep attentive panoptic model for prostate cancer detection using biparametric mri scans. In:International Conference on Medical Image Computing and Computer\u2010Assisted Intervention.Springer;2020:594\u2010604.","DOI":"10.1007\/978-3-030-59719-1_58"},{"key":"e_1_2_9_25_1","doi-asserted-by":"publisher","DOI":"10.1007\/s00330-021-08320-y"},{"key":"e_1_2_9_26_1","doi-asserted-by":"publisher","DOI":"10.3390\/cancers11121860"},{"key":"e_1_2_9_27_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2021.102288"},{"key":"e_1_2_9_28_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.juro.2016.07.084"},{"key":"e_1_2_9_29_1","doi-asserted-by":"publisher","DOI":"10.1002\/jmri.24629"},{"key":"e_1_2_9_30_1","doi-asserted-by":"publisher","DOI":"10.1117\/1.JMI.5.1.011004"},{"key":"e_1_2_9_31_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.compmedimag.2017.12.002"},{"key":"e_1_2_9_32_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.14337"},{"key":"e_1_2_9_33_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2020.101919"},{"key":"e_1_2_9_34_1","doi-asserted-by":"crossref","unstructured":"ShaoW BhattacharyaI SoerensenSJ et\u00a0al.Weakly supervised registration of prostate MRI and histopathology images. In:International Conference on Medical Image Computing and Computer\u2010Assisted Intervention.Springer;2021:98\u2010107.","DOI":"10.1007\/978-3-030-87202-1_10"},{"key":"e_1_2_9_35_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2021.101957"},{"key":"e_1_2_9_36_1","unstructured":"BultenW BalkenholM BelingaJ\u2010JA et\u00a0al.Artificial Intelligence Assistance Significantly Improves Gleason Grading of Prostate Biopsies by Pathologists. arXiv:2002.04500.2020."},{"key":"e_1_2_9_37_1","doi-asserted-by":"publisher","DOI":"10.1016\/S1470-2045(19)30739-9"},{"key":"e_1_2_9_38_1","doi-asserted-by":"crossref","unstructured":"RonnebergerO FischerP BroxT.U\u2010net: convolutional networks for biomedical image segmentation. In:International Conference on Medical image computing and computer\u2010assisted intervention.Springer;2015:234\u2010241.","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"e_1_2_9_39_1","doi-asserted-by":"publisher","DOI":"10.5489\/cuaj.4859"},{"key":"e_1_2_9_40_1","doi-asserted-by":"publisher","DOI":"10.5152\/dir.2014.13319"},{"key":"e_1_2_9_41_1","doi-asserted-by":"publisher","DOI":"10.1007\/s00330-020-07256-z"},{"key":"e_1_2_9_42_1","doi-asserted-by":"publisher","DOI":"10.1109\/TBME.2020.2993528"},{"key":"e_1_2_9_43_1","doi-asserted-by":"publisher","DOI":"10.1097\/JU.0000000000002090.03"},{"key":"e_1_2_9_44_1","doi-asserted-by":"publisher","DOI":"10.1109\/42.836373"},{"key":"e_1_2_9_45_1","doi-asserted-by":"crossref","unstructured":"ReinholdJC DeweyBE CarassA PrinceJL.Evaluating the impact of intensity normalization on MR image synthesis. In:Medical Imaging 2019: Image Processing. vol.10949.International Society for Optics and Photonics;2019:109493H.","DOI":"10.1117\/12.2513089"},{"key":"e_1_2_9_46_1","doi-asserted-by":"publisher","DOI":"10.1111\/his.13712"},{"key":"e_1_2_9_47_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.eururo.2019.02.033"},{"key":"e_1_2_9_48_1","doi-asserted-by":"publisher","DOI":"10.1148\/radiol.2019190938"},{"key":"e_1_2_9_49_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-021-01620-2"},{"key":"e_1_2_9_50_1","unstructured":"HosseinzadehM BrandP HuismanH.Effect of adding probabilistic zonal prior in deep learning\u2010based prostate cancer detection. arXiv:1907.12382.2019."}],"container-title":["Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.15777","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/mp.15777","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.15777","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,23]],"date-time":"2023-08-23T03:00:57Z","timestamp":1692759657000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/mp.15777"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,6,13]]},"references-count":49,"journal-issue":{"issue":"8","published-print":{"date-parts":[[2022,8]]}},"alternative-id":["10.1002\/mp.15777"],"URL":"https:\/\/doi.org\/10.1002\/mp.15777","archive":["Portico"],"relation":{},"ISSN":["0094-2405","2473-4209"],"issn-type":[{"value":"0094-2405","type":"print"},{"value":"2473-4209","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,6,13]]},"assertion":[{"value":"2021-12-01","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-05-10","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-06-13","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}