{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,16]],"date-time":"2026-10-16T13:12:18Z","timestamp":1792156338204,"version":"4.4.0"},"reference-count":32,"publisher":"Wiley","issue":"12","license":[{"start":{"date-parts":[[2025,11,23]],"date-time":"2025-11-23T00:00:00Z","timestamp":1763856000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["aapm.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["J Applied Clin Med Phys"],"published-print":{"date-parts":[[2025,12]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>Although bone mineral density (BMD) measurement using dual\u2010energy x\u2010ray absorptiometry (DXA) is the most common method of diagnosing osteoporosis, it is not widely used to screen patients. In this exploratory study, we developed a multi\u2010task learning model that predicts BMD from chest radiographs using clavicular features and supports network explainability.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      The proposed multi\u2010task learning model integrates segmentation and regression tasks by incorporating a regression branch into the U\u2010Net architecture in an end\u2010to\u2010end manner. A total of 1600 patients who underwent chest radiography and DXA of the lumbar vertebrae were included in this study. We compared the BMD predictive performance of the mean absolute error (MAE) and Pearson correlation coefficient (\n                      <jats:italic>R<\/jats:italic>\n                      value) between the proposed multi\u2010task learning model and the single\u2010task learning model, which was defined as the comparison model that only performed BMD prediction. Additionally, model performance for classifying osteoporosis, osteopenia, and normal bone status was evaluated via reclassification analysis based on the World Health Organization (WHO) criteria. Confusion matrices were generated, and classwise and macro\u2010averaged performance metrics were calculated. To confirm the rationale for the BMD predictions, we evaluated heat maps generated using the gradient\u2010weighted class activation mapping technique to determine whether the highlighted regions overlapped with the clavicle.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      The multi\u2010task learning model demonstrated superior predictive performance (MAE of 0.092\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      and\n                      <jats:italic>R<\/jats:italic>\n                      value of 0.769) compared with the single\u2010task learning model (MAE of 0.101\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      and\n                      <jats:italic>R<\/jats:italic>\n                      value of 0.724), a statistically significant (\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0&lt;\u00a00.001) difference in MAE. Bland\u2013Altman analysis showed that the multi\u2010task learning model had good agreement with narrower limits of agreement, although a bias was present (bias: \u22120.013\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      ; limits of agreement: \u22120.248 to 0.223\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      ). In contrast, the single\u2010task model showed slightly wider agreement limits (bias: \u22120.003\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      ; limits of agreement: \u22120.257 to 0.252\u00a0g\/cm\n                      <jats:sup>2<\/jats:sup>\n                      ). In the reclassification analysis based on the WHO criteria, the multi\u2010task learning model resulted in fewer misclassifications than the single\u2010task learning model. The macro\u2010averaged sensitivity, specificity, precision, and F1 score were 0.647, 0.826, 0.680, and 0.659, respectively, for the multi\u2010task model, compared with 0.597, 0.809, 0.660, and 0.616, respectively, for the single\u2010task model. The heat maps in the multi\u2010task learning model highlighted different regions compared with the single\u2010task model, the clavicular area.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>The proposed multi\u2010task learning model demonstrated the predictive rationale by focusing on the clavicle in chest radiographs, which is clinically relevant to BMD, and showed improved performance compared with the single\u2010task model.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1002\/acm2.70336","type":"journal-article","created":{"date-parts":[[2025,11,24]],"date-time":"2025-11-24T07:49:01Z","timestamp":1763970541000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":4,"title":["An exploratory study of explainable deep learning for predicting bone mineral density using clavicle features on chest radiographs: A multi\u2010task approach with regression and segmentation"],"prefix":"10.1002","volume":"26","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-8713-1845","authenticated-orcid":false,"given":"Yoshiyuki","family":"Iwao","sequence":"first","affiliation":[{"name":"Medical Technology Department Oita University Hospital  Yufu City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0002-0337-2005","authenticated-orcid":false,"given":"Kenshi","family":"Shiotsuki","sequence":"additional","affiliation":[{"name":"Medical Technology Department Oita University Hospital  Yufu City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2352-0538","authenticated-orcid":false,"given":"Fumio","family":"Hashimoto","sequence":"additional","affiliation":[{"name":"Graduate School of Science and Engineering Chiba University  Chiba Japan"},{"name":"J. Crayton Pruitt Family Department of Biomedical Engineering University of Florida  Gainesville USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Takahiro","family":"Ochiai","sequence":"additional","affiliation":[{"name":"Medical Technology Department Oita University Hospital  Yufu City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tsuneo","family":"Kagawa","sequence":"additional","affiliation":[{"name":"Division of Computer Science and Intelligent Systems Faculty of Science and Technology Oita University  Oita City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ryoichi","family":"Nagata","sequence":"additional","affiliation":[{"name":"Division of Computer Science and Intelligent Systems Faculty of Science and Technology Oita University  Oita City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-2159-2015","authenticated-orcid":false,"given":"Michihiro","family":"Eto","sequence":"additional","affiliation":[{"name":"Department of Health and Medical Sciences Nippon Bunri University  Oita City Oita Japan"},{"name":"Graduate School of Engineering Oita University  Oita City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8621-5592","authenticated-orcid":false,"given":"Yuji","family":"Hatanaka","sequence":"additional","affiliation":[{"name":"Division of Computer Science and Intelligent Systems Faculty of Science and Technology Oita University  Oita City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yukito","family":"Yoshida","sequence":"additional","affiliation":[{"name":"Medical Technology Department Oita University Hospital  Yufu City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6495-4821","authenticated-orcid":false,"given":"Yoshiki","family":"Asayama","sequence":"additional","affiliation":[{"name":"Department of Radiology Oita University Faculty of Medicine  Yufu City Oita Japan"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2025,11,23]]},"reference":[{"key":"e_1_2_11_2_1","doi-asserted-by":"publisher","DOI":"10.5152\/eurjrheum.2016.048"},{"key":"e_1_2_11_3_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.mcna.2021.05.016"},{"key":"e_1_2_11_4_1","doi-asserted-by":"publisher","DOI":"10.1007\/s00198\u2010004\u20101762\u20107"},{"key":"e_1_2_11_5_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11914\u2010005\u20100005\u2010y"},{"key":"e_1_2_11_6_1","doi-asserted-by":"publisher","DOI":"10.1080\/21556660.2019.1579728"},{"key":"e_1_2_11_7_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.bone.2020.115561"},{"key":"e_1_2_11_8_1","doi-asserted-by":"publisher","DOI":"10.3389\/fendo.2022.971877"},{"key":"e_1_2_11_9_1","doi-asserted-by":"publisher","DOI":"10.3390\/biom10111534"},{"key":"e_1_2_11_10_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41598\u2010021\u201099549\u20106"},{"key":"e_1_2_11_11_1","doi-asserted-by":"publisher","DOI":"10.1109\/EMBC.2018.8512755"},{"key":"e_1_2_11_12_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41598\u2010022\u201010150\u2010x"},{"key":"e_1_2_11_13_1","doi-asserted-by":"publisher","DOI":"10.1007\/s40618\u2010014\u20100074\u20109"},{"issue":"5","key":"e_1_2_11_14_1","first-page":"3","article-title":"Development of a fast screening method for osteoporosis using chest x\u2010ray images and machine learning","volume":"3","author":"Ohta Y","year":"2020","journal-title":"Can J Biomed Res and Tech"},{"key":"e_1_2_11_15_1","doi-asserted-by":"publisher","DOI":"10.1002\/jbmr.4477"},{"key":"e_1_2_11_16_1","doi-asserted-by":"publisher","DOI":"10.3390\/biomedicines10092323"},{"key":"e_1_2_11_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2022.3209648"},{"key":"e_1_2_11_18_1","doi-asserted-by":"publisher","DOI":"10.1007\/s12194\u2010023\u201000697\u20103"},{"key":"e_1_2_11_19_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2022.102470"},{"key":"e_1_2_11_20_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ejrad.2023.110787"},{"key":"e_1_2_11_21_1","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2020190043"},{"key":"e_1_2_11_22_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11263\u2010019\u201001228\u20107"},{"key":"e_1_2_11_23_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.jneumeth.2021.109098"},{"key":"e_1_2_11_24_1","doi-asserted-by":"publisher","DOI":"10.1093\/nsr\/nwx105"},{"key":"e_1_2_11_25_1","doi-asserted-by":"publisher","DOI":"10.1007\/978\u20103\u2010319\u201024574\u20104_28"},{"key":"e_1_2_11_26_1","doi-asserted-by":"publisher","DOI":"10.3389\/frai.2024.1423535"},{"key":"e_1_2_11_27_1","volume-title":"European guidelines on quality criteria for diagnostic radiographic images","author":"European Commission: Directorate\u2010General for Research and Innovation","year":"1996"},{"key":"e_1_2_11_28_1","doi-asserted-by":"publisher","DOI":"10.1186\/s12891\u2010022\u201005015\u20103"},{"key":"e_1_2_11_29_1","first-page":"1","volume-title":"Assessment of Fracture Risk and Its Application to Screening for Postmenopausal Osteoporosis","author":"World Health Organization","year":"1994"},{"key":"e_1_2_11_30_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-11149-6_1"},{"key":"e_1_2_11_31_1","doi-asserted-by":"publisher","DOI":"10.1148\/ryai.2021200267"},{"key":"e_1_2_11_32_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256\u2010021\u201000338\u20107"},{"key":"e_1_2_11_33_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42256\u2010022\u201000536\u2010x"}],"container-title":["Journal of Applied Clinical Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/acm2.70336","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,31]],"date-time":"2025-12-31T11:13:33Z","timestamp":1767179613000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/acm2.70336"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,23]]},"references-count":32,"journal-issue":{"issue":"12","published-print":{"date-parts":[[2025,12]]}},"alternative-id":["10.1002\/acm2.70336"],"URL":"https:\/\/doi.org\/10.1002\/acm2.70336","archive":["Portico"],"relation":{},"ISSN":["1526-9914","1526-9914"],"issn-type":[{"value":"1526-9914","type":"print"},{"value":"1526-9914","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,11,23]]},"assertion":[{"value":"2024-12-10","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-09-30","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-11-23","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70336"}}