{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,11]],"date-time":"2026-09-11T20:54:56Z","timestamp":1789160096173,"version":"build-2803163510"},"reference-count":25,"publisher":"Wiley","issue":"1","license":[{"start":{"date-parts":[[2023,11,1]],"date-time":"2023-11-01T00:00:00Z","timestamp":1698796800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"content-domain":{"domain":["aapm.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Medical Physics"],"published-print":{"date-parts":[[2024,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Accurate 3D semantic segmentation models are essential for many clinical applications. To train a model for 3D segmentation, voxel\u2010level annotation is necessary, which is expensive to obtain due to laborious work and privacy protection. To accurately annotate 3D medical data, such as MRI, a common practice is to annotate the volumetric data in a slice\u2010by\u2010slice contouring way along principal\u00a0axes.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Purpose<\/jats:title>\n                    <jats:p>In order to reduce the annotation effort in slices, weakly supervised learning with a bounding box (Bbox) was proposed to leverage the discriminating information via a tightness prior assumption. Nevertheless, this method requests accurate and tight Bboxes, which will significantly drop the performance when tightness is not held, that is when a relaxed Bbox is applied. Therefore, there is a need to train a stable model based on relaxed Bbox\u00a0annotation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This paper presents a mixed\u2010supervised training strategy to reduce the annotation effort for 3D segmentation tasks. In the proposed approach, a fully annotated contour is only required for a single slice of the volume. In contrast, the rest of the slices with targets are annotated with relaxed Bboxes. This mixed\u2010supervised method adopts fully supervised learning, relaxed Bbox prior, and contrastive learning during the training, which ensures the network exploits the discriminative information of the training volumes properly. The proposed method was evaluated on two public 3D medical imaging datasets (MRI prostate dataset and Vestibular Schwannoma [VS] dataset).<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The proposed method obtained a high segmentation Dice score of 85.3% on an MRI prostate dataset and 83.3% on a VS dataset with relaxed Bbox annotation, which are close to a fully supervised model. Moreover, with the same relaxed Bbox annotations, the proposed method outperforms the state\u2010of\u2010the\u2010art methods. More importantly, the model performance is stable when the accuracy of Bbox annotation\u00a0varies.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>The presented study proposes a method based on a mixed\u2010supervised learning method in 3D medical imaging. The benefit will be stable segmentation of the target in 3D images with low accurate annotation requirement, which leads to easier model training on large\u2010scale\u00a0datasets.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1002\/mp.16816","type":"journal-article","created":{"date-parts":[[2023,11,1]],"date-time":"2023-11-01T09:41:29Z","timestamp":1698831689000},"page":"167-178","update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Light mixed\u2010supervised segmentation for 3D medical image data"],"prefix":"10.1002","volume":"51","author":[{"given":"Hongxu","family":"Yang","sequence":"first","affiliation":[{"name":"GE Healthcare Eindhoven The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tao","family":"Tan","sequence":"additional","affiliation":[{"name":"GE Healthcare Eindhoven The Netherlands"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pal","family":"Tegzes","sequence":"additional","affiliation":[{"name":"GE Healthcare Budapest Hungary"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaomeng","family":"Dong","sequence":"additional","affiliation":[{"name":"GE Healthcare San Ramon USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rajesh","family":"Tamada","sequence":"additional","affiliation":[{"name":"GE Healthcare Bangalore India"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lehel","family":"Ferenczi","sequence":"additional","affiliation":[{"name":"GE Healthcare Budapest Hungary"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Gopal","family":"Avinash","sequence":"additional","affiliation":[{"name":"GE Healthcare San Ramon USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2023,11]]},"reference":[{"key":"e_1_2_10_2_1","first-page":"365","volume-title":"Medical Imaging with Deep Learning","author":"Kervadec H","year":"2020"},{"key":"e_1_2_10_3_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-87196-3_49"},{"key":"e_1_2_10_4_1","first-page":"615","volume-title":"International Conference on Medical Image Computing and Computer\u2010Assisted Intervention","author":"Dorent R","year":"2021"},{"key":"e_1_2_10_5_1","first-page":"605","volume-title":"International Conference on Medical Image Computing and Computer\u2010Assisted Intervention","author":"Yu L","year":"2019"},{"key":"e_1_2_10_6_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2020.101693"},{"key":"e_1_2_10_7_1","article-title":"Weakly supervised instance segmentation using the bounding box tightness prior","volume":"32","author":"Hsu C\u2010C","year":"2019","journal-title":"Adv Neural Inf Process Syst"},{"key":"e_1_2_10_8_1","doi-asserted-by":"crossref","unstructured":"LaiX TianZ JiangL et\u00a0al.Semi\u2010supervised semantic segmentation with directional context\u2010aware consistency. In:Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition.IEEE;2021:1205\u20101214.","DOI":"10.1109\/CVPR46437.2021.00126"},{"key":"e_1_2_10_9_1","doi-asserted-by":"crossref","unstructured":"ShahMP MerchantS AwateSP.MS\u2010Net: mixed\u2010supervision fully\u2010convolutional networks for full\u2010resolution segmentation. In:Medical Image Computing and Computer Assisted Intervention\u2013MICCAI 2018: 21st International Conference Granada Spain September 16\u201020 2018 Proceedings Part IV 11.Springer;2018:379\u2010387.","DOI":"10.1007\/978-3-030-00937-3_44"},{"key":"e_1_2_10_10_1","doi-asserted-by":"crossref","unstructured":"WangD LiM Ben\u2010ShlomoN et\u00a0al.Mixed\u2010supervised dual\u2010network for medical image segmentation. In:Medical Image Computing and Computer Assisted Intervention\u2013MICCAI 2019: 22nd International Conference Shenzhen China October 13\u201317 2019 Proceedings Part II 22.Springer;2019:192\u2010200.","DOI":"10.1007\/978-3-030-32245-8_22"},{"key":"e_1_2_10_11_1","doi-asserted-by":"crossref","unstructured":"DolzJ DesrosiersC AyedIB.Teach me to segment with mixed supervision: Confident students become masters. In:Information Processing in Medical Imaging: 27th International Conference IPMI 2021 Virtual Event June 28\u2013June 30 2021 Proceedings 27.Springer;2021:517\u2010529.","DOI":"10.1007\/978-3-030-78191-0_40"},{"key":"e_1_2_10_12_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2022.102670"},{"key":"e_1_2_10_13_1","doi-asserted-by":"publisher","DOI":"10.1109\/JBHI.2021.3101872"},{"key":"e_1_2_10_14_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2016.2621185"},{"key":"e_1_2_10_15_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.compmedimag.2022.102037"},{"key":"e_1_2_10_16_1","first-page":"1","article-title":"A survey of unsupervised learning in medical image registration","volume":"2","author":"Song X","year":"2022","journal-title":"International Journal of Health Systems and Translational Medicine (IJHSTM)"},{"key":"e_1_2_10_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.319"},{"key":"e_1_2_10_18_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11548-019-01958-6"},{"key":"e_1_2_10_19_1","article-title":"Efficient inference in fully connected CRFs with gaussian edge potentials","volume":"24","author":"Kr\u00e4henb\u00fchl P","year":"2011","journal-title":"Adv Neural Inf Process Syst"},{"key":"e_1_2_10_20_1","doi-asserted-by":"crossref","unstructured":"LuoX HuM LiaoW et\u00a0al.Scribble\u2010supervised medical image segmentation via dual\u2010branch network and dynamically mixed pseudo labels supervision. In:Medical Image Computing and Computer Assisted Intervention\u2013MICCAI 2022: 25th International Conference Singapore September 18\u201322 2022 Proceedings Part I.Springer;2022:528\u2010538.","DOI":"10.1007\/978-3-031-16431-6_50"},{"key":"e_1_2_10_21_1","first-page":"12546","article-title":"Contrastive learning of global and local features for medical image segmentation with limited annotations","volume":"33","author":"Chaitanya K","year":"2020","journal-title":"Adv Neural Inf Process Syst"},{"key":"e_1_2_10_22_1","article-title":"Mean teachers are better role models: weight\u2010averaged consistency targets improve semi\u2010supervised deep learning results","volume":"30","author":"Tarvainen A","year":"2017","journal-title":"Adv Neural Inf Process Syst"},{"key":"e_1_2_10_23_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2013.12.002"},{"key":"e_1_2_10_24_1","doi-asserted-by":"publisher","DOI":"10.3171\/2019.9.JNS191949"},{"key":"e_1_2_10_25_1","unstructured":"KingmaDP BaJ.Adam: a method for stochastic optimization. In:BengioY LeCunY eds.3rd International Conference on Learning Representations. Conference Track Proceedings;2015.https:\/\/dblp.org\/rec\/journals\/corr\/KingmaB14.bib bibsource = dblp computer science bibliography https:\/\/dblp.org"},{"key":"e_1_2_10_26_1","unstructured":"IsenseeF PetersenJ KleinA et\u00a0al.nnU\u2010Net: self\u2010adapting framework for U\u2010Net\u2010based medical image segmentation. arXiv preprint arXiv:1809.10486.2018."}],"container-title":["Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.16816","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,1,18]],"date-time":"2024-01-18T15:53:41Z","timestamp":1705593221000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/mp.16816"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,11]]},"references-count":25,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2024,1]]}},"alternative-id":["10.1002\/mp.16816"],"URL":"https:\/\/doi.org\/10.1002\/mp.16816","archive":["Portico"],"relation":{},"ISSN":["0094-2405","2473-4209"],"issn-type":[{"value":"0094-2405","type":"print"},{"value":"2473-4209","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,11]]},"assertion":[{"value":"2023-01-13","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-10-16","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-11-01","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}