{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,8]],"date-time":"2026-05-08T17:09:06Z","timestamp":1778260146699,"version":"3.51.4"},"reference-count":47,"publisher":"Wiley","issue":"1","license":[{"start":{"date-parts":[[2025,12,31]],"date-time":"2025-12-31T00:00:00Z","timestamp":1767139200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"},{"start":{"date-parts":[[2025,12,31]],"date-time":"2025-12-31T00:00:00Z","timestamp":1767139200000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"content-domain":{"domain":["aapm.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Medical Physics"],"published-print":{"date-parts":[[2026,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>With the increasing number of diabetic patients, the rapid and accurate diagnosis of early diabetic\u00a0retinopathy becomes crucial. However, diabetic retinal lesions are challenging to label since the identification of disease depends on accessing multiple lesion regions in images, which requires specialists to make detailed judgments and label them, resulting in an extremely costly and time\u2010consuming.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>To reduce costs, we propose diabetic\u00a0retinopathy\u00a0detection\u00a0network (DRD\u2010Net), an improved weakly supervised object detection model based on adversarial complementary erasure learning (ACoL) framework, designed for diabetic retinopathy detection. DRD\u2010Net enhances small lesion localization while relying only on image\u2010level labels.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>\n                      DRD includes an improved EfficientNet\u2010B0 network, which leverages compressed network structure with parallel downsampling and the efficient channel attention (ECA) module for feature extraction from fundus images. A multi\u2010scale parallel attention module (MPA) is designed and combines with adversarial complementary erasure learning to enhance classification and localization of small lesion accross multi\u2010scale features. For data processing, we cropped and re\u2010annotated three datasets into 35,828 lesion patches (224\u00a0\u00d7\u00a0224 pixels) to solve the problem of information loss in high\u2010resolution fundus images. The dataset is partitioned into training (25,079, 70%), validation (7166, 20%), and test (3583, 10%) sets. Benchmark models include CNN\u2010based methods (CAM, ACoL, SLT\u2010Net, etc.) and Transformer\u2010based approaches (TRT, SAT, etc.). Performance is evaluated using Top\u20101\/Top\u20105 classification accuracy (Top\u20101\/Top\u20105 Cls), Top\u20101\/Top\u20105 localization accuracy (Top\u20101\/Top\u20105 Loc), and ground\u2010truth known localization accuracy (GT\u2010Known Loc). Statistical analyses employs paired\n                      <jats:italic>t<\/jats:italic>\n                      \u2010tests with Holm\u2010Bonferroni correction for multiple comparisons, Cohen's d for effect size, and a significance level of \u03b1\u00a0=\u00a00.05.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>\n                      Experiments verify that the performance of DRD\u2010Net is better than the state\u2010of\u2010the\u2010art methods, achieving 82.41%, 76.94%, and 86.05% in Top\u20101 Cls, Top\u20101 Loc and GT\u2010Known Loc, respectively. Compared to top\u2010performancing baselines, gains are 1.64% (Cohen's d\u00a0=\u00a01.016,\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.0318), 3.16% (\n                      <jats:italic>d<\/jats:italic>\n                      \u00a0=\u00a01.377,\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.0090), and 0.96% (\n                      <jats:italic>d<\/jats:italic>\n                      \u00a0=\u00a00.85,\n                      <jats:italic>p<\/jats:italic>\n                      \u00a0=\u00a00.0481), all significant at \u03b1\u00a0=\u00a00.05.\n                    <\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Experiments confirm that DRD\u2010Net has good feasibility to accurately and comprehensively identify DR lesions. This suggests that it could potentially enhance clinical screening efficiency and promote further development in diabetic retinopathy detection.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1002\/mp.70264","type":"journal-article","created":{"date-parts":[[2025,12,31]],"date-time":"2025-12-31T11:12:36Z","timestamp":1767179556000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Weakly supervised object detection network for diabetic retinopathy"],"prefix":"10.1002","volume":"53","author":[{"given":"Xulin","family":"Zong","sequence":"first","affiliation":[{"name":"Faculty of Information Science and Engineering Ocean University of China  Qingdao China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Bingxue","family":"Liang","sequence":"additional","affiliation":[{"name":"College of Information Science and Technology Qingdao University of Science and Technology  Qingdao China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yuhua","family":"Qin","sequence":"additional","affiliation":[{"name":"College of Information Science and Technology Qingdao University of Science and Technology  Qingdao China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xiangqian","family":"Ding","sequence":"additional","affiliation":[{"name":"Faculty of Information Science and Engineering Ocean University of China  Qingdao China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenjing","family":"Wang","sequence":"additional","affiliation":[{"name":"College of Information Science and Technology Qingdao University of Science and Technology  Qingdao China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"311","published-online":{"date-parts":[[2025,12,31]]},"reference":[{"key":"e_1_2_8_2_1","doi-asserted-by":"publisher","DOI":"10.1080\/07853890.2023.2258149"},{"key":"e_1_2_8_3_1","volume-title":"IDF Diabetes Atlas. International Diabetes Federation","author":"Atlas D","year":"2021"},{"key":"e_1_2_8_4_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2107.00115"},{"key":"e_1_2_8_5_1","doi-asserted-by":"publisher","DOI":"10.3389\/fpubh.2022.971943"},{"key":"e_1_2_8_6_1","doi-asserted-by":"crossref","unstructured":"StoianAP.Type 2 diabetes \u2010 from pathophysiology to cyber systems.Croatia: IntechOpen.2021; doi:10.5772\/intechopen.87312","DOI":"10.5772\/intechopen.87312"},{"key":"e_1_2_8_7_1","doi-asserted-by":"publisher","DOI":"10.3390\/diabetology3010011"},{"key":"e_1_2_8_8_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.matpr.2017.12.372"},{"key":"e_1_2_8_9_1","doi-asserted-by":"publisher","DOI":"10.1016\/J.BBE.2021.04.005"},{"key":"e_1_2_8_10_1","doi-asserted-by":"publisher","DOI":"10.3390\/bdcc7010025"},{"key":"e_1_2_8_11_1","doi-asserted-by":"publisher","DOI":"10.3390\/s21113865"},{"key":"e_1_2_8_12_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.compeleceng.2024.109746"},{"key":"e_1_2_8_13_1","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.7405"},{"key":"e_1_2_8_14_1","doi-asserted-by":"publisher","DOI":"10.1002\/ima.22769"},{"key":"e_1_2_8_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2022.3157632"},{"key":"e_1_2_8_16_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.compeleceng.2024.109243"},{"key":"e_1_2_8_17_1","doi-asserted-by":"publisher","DOI":"10.11834\/jig.220854"},{"key":"e_1_2_8_18_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.319"},{"key":"e_1_2_8_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00144"},{"key":"e_1_2_8_20_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2002.11359"},{"key":"e_1_2_8_21_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.00732"},{"key":"e_1_2_8_22_1","doi-asserted-by":"crossref","unstructured":"ChenZ WangC WangY et\u00a0al.LCTR: on awakening the local continuity of transformer for weakly supervised object localization.Proc AAAI Conf Artif Intell.2022;410\u2010418. doi:10.1609\/aaai.v36i1.19995","DOI":"10.1609\/aaai.v36i1.19918"},{"key":"e_1_2_8_23_1","unstructured":"SuH YeY ChenZ et\u00a0al.Re\u2010attention transformer for weakly supervised object localization.Proc Br Mach Vis Conf.2022; doi:10.48550\/arXiv.2208.01838"},{"key":"e_1_2_8_24_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2025.113012"},{"key":"e_1_2_8_25_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1705.07999"},{"key":"e_1_2_8_26_1","doi-asserted-by":"publisher","DOI":"10.1109\/JTEHM.2023.3286423"},{"key":"e_1_2_8_27_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2402.11985"},{"key":"e_1_2_8_28_1","unstructured":"SongJ YunS YoonS et\u00a0al.EP\u2010SAM: Weakly supervised histopathology segmentation via enhanced prompt with segment anything. arxiv preprint arxiv: 2410.13621 2024. doi:10.48550\/arXiv.2410.13621"},{"key":"e_1_2_8_29_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2410.19332"},{"key":"e_1_2_8_30_1","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2018.2816003"},{"key":"e_1_2_8_31_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2018.2885376"},{"key":"e_1_2_8_32_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.compbiomed.2022.105341"},{"key":"e_1_2_8_33_1","doi-asserted-by":"publisher","DOI":"10.3390\/bioengineering10091100"},{"key":"e_1_2_8_34_1","unstructured":"TanM LeQ V.EfficientNet: rethinking model scaling for convolutional neural networks.Proc Int Conf Mach Learn.2019;6105\u20106114. doi:10.1109\/ICML.2019.00631"},{"key":"e_1_2_8_35_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.01155"},{"key":"e_1_2_8_36_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2019.101561"},{"key":"e_1_2_8_37_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ins.2019.06.011"},{"key":"e_1_2_8_38_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00745"},{"key":"e_1_2_8_39_1","unstructured":"HowardA G ZhuM ChenB et\u00a0al.MobileNets: efficient convolutional neural networks for mobile vision applications. ArXiv preprint arXiv:1704.04861 2017. doi:10.48550\/arXiv.1704.04861"},{"key":"e_1_2_8_40_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1312.4400"},{"key":"e_1_2_8_41_1","doi-asserted-by":"publisher","DOI":"10.5555\/2627435.2670313"},{"key":"e_1_2_8_42_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01157"},{"key":"e_1_2_8_43_1","unstructured":"WooS ParkJ LeeJ et\u00a0al.CBAM: convolutional block attention module.Proc Eue Conf Comput Vis Munich.2018;3\u20109. doi:10.1007\/978\u20103\u2010030\u201001234\u20102_1"},{"key":"e_1_2_8_44_1","first-page":"65","article-title":"A simple sequentially rejective multiple test procedure","volume":"6","author":"Holm S","year":"1979","journal-title":"Scand J Stat"},{"key":"e_1_2_8_45_1","doi-asserted-by":"publisher","DOI":"10.1525\/ae.2004.31.4.475"},{"key":"e_1_2_8_46_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00232"},{"key":"e_1_2_8_47_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2103.14862"},{"key":"e_1_2_8_48_1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2303.10438"}],"container-title":["Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.70264","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/mp.70264","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.70264","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,19]],"date-time":"2026-03-19T01:55:39Z","timestamp":1773885339000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/mp.70264"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,31]]},"references-count":47,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2026,1]]}},"alternative-id":["10.1002\/mp.70264"],"URL":"https:\/\/doi.org\/10.1002\/mp.70264","archive":["Portico"],"relation":{},"ISSN":["0094-2405","2473-4209"],"issn-type":[{"value":"0094-2405","type":"print"},{"value":"2473-4209","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,31]]},"assertion":[{"value":"2025-08-22","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-12-18","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-12-31","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70264"}}