{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T07:21:15Z","timestamp":1787556075378,"version":"build-2736575974"},"reference-count":37,"publisher":"IOP Publishing","issue":"16","license":[{"start":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T00:00:00Z","timestamp":1787529600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/publishingsupport.iopscience.iop.org\/iop-standard\/v1"},{"start":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T00:00:00Z","timestamp":1787529600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"},{"start":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T00:00:00Z","timestamp":1787529600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/publishingsupport.iopscience.iop.org\/questions\/alternative-author-rights-policies\/"},{"start":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T00:00:00Z","timestamp":1787529600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/publishingsupport.iopscience.iop.org\/questions\/alternative-author-rights-policies\/"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Phys. Med. Biol."],"published-print":{"date-parts":[[2026,8,28]]},"abstract":"Abstract<\/jats:title>\n \n Objective.<\/jats:italic>\n Breast ultrasound lesion segmentation remains challenging because heterogeneous echogenicity, speckle, posterior shadowing, and weak boundaries impair lesion-margin delineation.\n Approach.<\/jats:italic>\n We evaluate a CMU-Net-based implementation, termed RTCMUNet, in which the proposed ML-Retinex block modulates skip features before the inherited CMU-Net multi-scale attention gate. ML-Retinex combines learnable multi-scale projections, fixed Gaussian low-pass context estimates, clipped feature-ratio cues, and residual gating. It operates on learned skip features and neither estimates nor corrects scanner time-gain compensation. Nine models were evaluated on BUSI under a common three-run internal-validation protocol whose mutually disjoint validation subsets formed a pooled set of 389 cases. We assessed seven case-level metrics, paired Wilcoxon tests against CMU-Net with Holm correction, exploratory design-stage analyses, diagnostic controls, a secondary historical BUS\/BUSI\/TNSCUI benchmark, and a synthetic brightness-gradient stress proxy.\n Main results.<\/jats:italic>\n RTCMUNet achieved Dice\n \n \n \n <\/jats:tex-math>\n \n \n 0.826<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 0.220<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n , IoU\n \n \n \n <\/jats:tex-math>\n \n \n 0.748<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 0.240<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n , HD95\n \n \n \n <\/jats:tex-math>\n \n \n 20.65<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 32.94<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n \u2009px, and ASSD\n \n \n \n <\/jats:tex-math>\n \n \n 7.14<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 14.04<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n \u2009px. Sensitivity was\n \n \n \n <\/jats:tex-math>\n \n \n 0.848<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 0.212<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n , precision\n \n \n \n <\/jats:tex-math>\n \n \n 0.838<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 0.239<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n , and BF1\n \n \n \n <\/jats:tex-math>\n \n \n 0.570<\/mml:mn>\n \n \u00b1<\/mml:mo>\n <\/mml:mrow>\n 0.291<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n . Relative to CMU-Net, Dice and IoU increased by\n \n \n \n <\/jats:tex-math>\n \n \n 0.017<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n and\n \n \n \n <\/jats:tex-math>\n \n \n 0.021<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n , while HD95 and ASSD decreased by\n \n \n \n <\/jats:tex-math>\n \n \n 3.11<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n and\n \n \n \n <\/jats:tex-math>\n \n \n 2.23<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n \u2009px. All seven paired tests remained significant after Holm correction (largest adjusted\n \n \n \n <\/jats:tex-math>\n \n \n p<\/mml:mi>\n =<\/mml:mo>\n 0.0104<\/mml:mn>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n ). RTCMUNet added\n \n \n \n <\/jats:tex-math>\n \n \n 5.9<\/mml:mn>\n %<\/mml:mi>\n <\/mml:mrow>\n <\/mml:math>\n <\/jats:inline-formula>\n parameters and maintained 42 frames per second.\n Significance.<\/jats:italic>\n Within retrospective internal validation, and without scanner-level gain data or an independent test set, the results support further evaluation of ML-Retinex as a compact feature-modulation strategy in the tested CMU-Net\/BUSI setting.\n <\/jats:p>","DOI":"10.1088\/1361-6560\/ae97d2","type":"journal-article","created":{"date-parts":[[2026,8,10]],"date-time":"2026-08-10T22:55:17Z","timestamp":1786402517000},"page":"165018","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["RTCMUNet: retinex-inspired feature modulation for breast ultrasound lesion segmentation"],"prefix":"10.1088","volume":"71","author":[{"ORCID":"https:\/\/orcid.org\/0009-0009-3830-7378","authenticated-orcid":true,"given":"Lin","family":"Ma","sequence":"first","affiliation":[{"name":"University of Shanghai for Science and Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9630-5516","authenticated-orcid":true,"given":"Wenjie","family":"Cai","sequence":"additional","affiliation":[{"name":"University of Shanghai for Science and Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"266","published-online":{"date-parts":[[2026,8,24]]},"reference":[{"key":"pmbae97d2bib1","doi-asserted-by":"publisher","DOI":"10.1016\/j.dib.2019.104863","type":"journal-article","article-title":"Dataset of breast ultrasound images","volume":"28","author":"Al-Dhabyani","year":"2020","journal-title":"Data Br."},{"key":"pmbae97d2bib2","first-page":"pp 17302","type":"conference-proceedings","article-title":"EfficientViT: lightweight multi-scale attention for high-resolution dense prediction","author":"Cai","year":"2023a"},{"key":"pmbae97d2bib3","doi-asserted-by":"publisher","DOI":"10.1016\/j.measurement.2026.121187","type":"journal-article","article-title":"ALEFB: adaptive lightweight enhanced block for measurement-oriented standard section identification in second-trimester fetal ultrasound","volume":"273","author":"Cai","year":"2026","journal-title":"Measurement"},{"key":"pmbae97d2bib4","first-page":"pp 12504","type":"conference-proceedings","article-title":"Retinexformer: one-stage Retinex-based transformer for low-light image enhancement","author":"Cai","year":"2023b"},{"key":"pmbae97d2bib5","first-page":"pp 205","type":"book","article-title":"Swin-Unet: Unet-like pure transformer for medical image segmentation","volume":"vol 13803","author":"Cao","year":"2022"},{"key":"pmbae97d2bib6","article-title":"TransUNet: transformers make strong encoders for medical image segmentation","author":"Chen","year":"2021","type":"preprint"},{"key":"pmbae97d2bib7","first-page":"pp 139","type":"book","article-title":"TN-SCUI 2020: Thyroid nodule segmentation and classification in ultrasound images challenge","author":"Guo","year":"2021"},{"key":"pmbae97d2bib8","first-page":"pp 1055","type":"conference-proceedings","article-title":"UNet 3+: a full-scale connected UNet for medical image segmentation","author":"Huang","year":"2020"},{"key":"pmbae97d2bib9","doi-asserted-by":"publisher","first-page":"5631","DOI":"10.1109\/JBHI.2025.3546345","type":"journal-article","article-title":"EMGANet: edge-aware multi-scale group-mix attention network for breast cancer ultrasound image segmentation","volume":"29","author":"Huang","year":"2025","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"pmbae97d2bib10","doi-asserted-by":"publisher","DOI":"10.1155\/2018\/5137904","type":"journal-article","article-title":"Machine learning in ultrasound computer-aided diagnostic systems: a survey","volume":"2018","author":"Huang","year":"2018","journal-title":"BioMed Res. Int."},{"key":"pmbae97d2bib11","doi-asserted-by":"publisher","first-page":"965","DOI":"10.1109\/83.597272","type":"journal-article","article-title":"A multiscale Retinex for bridging the gap between color images and the human observation of scenes","volume":"6","author":"Jobson","year":"1997","journal-title":"IEEE Trans. Image Process."},{"key":"pmbae97d2bib12","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1364\/JOSA.61.000001","type":"journal-article","article-title":"Lightness and Retinex theory","volume":"61","author":"Land","year":"1971","journal-title":"J. Opt. Soc. Am."},{"key":"pmbae97d2bib13","article-title":"MobileViT: light-weight, general-purpose and mobile-friendly vision transformer","author":"Mehta","year":"2022","type":"conference-proceedings"},{"key":"pmbae97d2bib14","article-title":"Attention U-Net: learning where to look for the pancreas","author":"Oktay","year":"2018","type":"preprint"},{"key":"pmbae97d2bib15","first-page":"pp 1053","type":"conference-proceedings","article-title":"MK-UNet: multi-kernel lightweight CNN for medical image segmentation","author":"Rahman","year":"2025"},{"key":"pmbae97d2bib16","first-page":"pp 234","type":"conference-proceedings","article-title":"U-Net: convolutional networks for biomedical image segmentation","author":"Ronneberger","year":"2015"},{"key":"pmbae97d2bib17","first-page":"pp 481","type":"book","article-title":"EGE-UNet: an Efficient Group Enhanced UNet for Skin Lesion Segmentation","author":"Ruan","year":"2023"},{"key":"pmbae97d2bib18","doi-asserted-by":"publisher","DOI":"10.1016\/j.imavis.2024.105149","type":"journal-article","article-title":"Nighttime image semantic segmentation with retinex theory","volume":"148","author":"Sun","year":"2024","journal-title":"Image Vis. Comput."},{"key":"pmbae97d2bib19","first-page":"pp 1","type":"conference-proceedings","article-title":"CMU-Net: a strong ConvMixer-based medical ultrasound image segmentation network","author":"Tang","year":"2023"},{"key":"pmbae97d2bib20","article-title":"Patches are all you need?","author":"Trockman","year":"2022","type":"conference-proceedings"},{"key":"pmbae97d2bib21","first-page":"pp 36","type":"conference-proceedings","article-title":"Medical transformer: gated axial-attention for medical image segmentation","author":"Valanarasu","year":"2021"},{"key":"pmbae97d2bib22","first-page":"pp 23","type":"conference-proceedings","article-title":"UNeXt: MLP-based rapid medical image segmentation network","author":"Valanarasu","year":"2022"},{"key":"pmbae97d2bib23","first-page":"pp 15909","type":"conference-proceedings","article-title":"RepViT: revisiting mobile CNN from ViT perspective","author":"Wang","year":"2024"},{"key":"pmbae97d2bib24","first-page":"pp 2441","type":"conference-proceedings","article-title":"UCTransNet: rethinking the skip connections in U-Net from a channel-wise perspective with transformer","volume":"vol 36","author":"Wang","year":"2022"},{"key":"pmbae97d2bib25","article-title":"Deep retinex decomposition for low-light enhancement","author":"Wei","year":"2018","type":"conference-proceedings"},{"key":"pmbae97d2bib26","doi-asserted-by":"publisher","first-page":"340","DOI":"10.1016\/j.patcog.2018.02.012","type":"journal-article","article-title":"Automatic breast ultrasound image segmentation: a survey","volume":"79","author":"Xian","year":"2018","journal-title":"Pattern Recognit."},{"key":"pmbae97d2bib27","doi-asserted-by":"publisher","first-page":"4","DOI":"10.1109\/TMI.2022.3192303","type":"journal-article","article-title":"Unsupervised domain adaptation for medical image segmentation by disentanglement learning and self-training","volume":"43","author":"Xie","year":"2024","journal-title":"IEEE Trans. Med. Imaging"},{"key":"pmbae97d2bib28","article-title":"SAM2-UNet: segment anything 2 makes strong encoder for natural and medical image segmentation","author":"Xiong","year":"2024","type":"preprint"},{"key":"pmbae97d2bib29","doi-asserted-by":"publisher","DOI":"10.1016\/j.dsp.2025.105264","type":"journal-article","article-title":"AO-TransUNet: a multi-attention optimization network for COVID-19 and medical image segmentation","volume":"164","author":"Yang","year":"2025","journal-title":"Digit. Signal Process."},{"key":"pmbae97d2bib30","article-title":"SegDINO: an efficient design for medical and natural image segmentation with DINO-V3","author":"Yang","year":"2025","type":"preprint"},{"key":"pmbae97d2bib31","doi-asserted-by":"publisher","first-page":"1218","DOI":"10.1109\/JBHI.2017.2731873","type":"journal-article","article-title":"Automated breast ultrasound lesions detection using convolutional neural networks","volume":"22","author":"Yap","year":"2018","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"pmbae97d2bib32","first-page":"pp 130","type":"book","article-title":"MI-SegNet: mutual information-based US segmentation for unseen domain generalization","volume":"vol 14223","author":"Yuan","year":"2023"},{"key":"pmbae97d2bib33","first-page":"pp 1632","type":"conference-proceedings","article-title":"Kindling the darkness: a practical low-light image enhancer","author":"Zhang","year":"2019"},{"key":"pmbae97d2bib34","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.3715942","type":"dataset","article-title":"Thyroid nodule segmentation and classification in ultrasound images","author":"Zhou","year":"2020","unstructured":"Zhou J Jia X Ni D Noble J A Huang R Tan T Van M T 2020 Thyroid nodule segmentation and classification in ultrasound images Zenodo 10.5281\/zenodo.3715942"},{"key":"pmbae97d2bib35","first-page":"pp 3","type":"conference-proceedings","article-title":"UNet++: a nested U-Net architecture for medical image segmentation","author":"Zhou","year":"2018"},{"key":"pmbae97d2bib36","doi-asserted-by":"publisher","first-page":"2258","DOI":"10.1016\/j.ultrasmedbio.2025.08.014","type":"journal-article","article-title":"Contrast and gain-aware attention: a plug-and-play feature fusion attention module for torso region fetal plane identification","volume":"51","author":"Zhu","year":"2025a","journal-title":"Ultrasound Med. Biol."},{"key":"pmbae97d2bib37","doi-asserted-by":"publisher","first-page":"12508","DOI":"10.1109\/TCSVT.2025.3588395","type":"journal-article","article-title":"Singular pooling: a spectral pooling paradigm for second-trimester prenatal level II ultrasound standard fetal plane identification","volume":"35","author":"Zhu","year":"2025b","journal-title":"IEEE Trans. Circuits Syst. Video Technol."}],"container-title":["Physics in Medicine & Biology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,24]],"date-time":"2026-08-24T06:49:45Z","timestamp":1787554185000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/1361-6560\/ae97d2"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,24]]},"references-count":37,"journal-issue":{"issue":"16","published-online":{"date-parts":[[2026,8,24]]},"published-print":{"date-parts":[[2026,8,28]]}},"URL":"https:\/\/doi.org\/10.1088\/1361-6560\/ae97d2","relation":{},"ISSN":["0031-9155","1361-6560"],"issn-type":[{"value":"0031-9155","type":"print"},{"value":"1361-6560","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,24]]},"assertion":[{"value":"RTCMUNet: retinex-inspired feature modulation for breast ultrasound lesion segmentation","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 2026 Institute of Physics and Engineering in Medicine. All rights, including for text and data mining, AI training, and similar technologies, are reserved.","name":"copyright_information","label":"Copyright Information"},{"value":"2026-04-22","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2026-08-10","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2026-08-24","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}