{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T06:18:45Z","timestamp":1790749125854,"version":"4.1.0"},"reference-count":41,"publisher":"Wiley","issue":"6","license":[{"start":{"date-parts":[[2019,5,6]],"date-time":"2019-05-06T00:00:00Z","timestamp":1557100800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"funder":[{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R01CA230278"],"award-info":[{"award-number":["R01CA230278"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R01CA188300"],"award-info":[{"award-number":["R01CA188300"]}],"id":[{"id":"10.13039\/100000002","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":[[2019,6]]},"abstract":"Purpose<\/jats:title>Image\u2010guided radiotherapy provides images not only for patient positioning but also for online adaptive radiotherapy. Accurate delineation of organs\u2010at\u2010risk (OAR<\/jats:styled-content>s) on Head and Neck (H&N) CT<\/jats:styled-content> and MR<\/jats:styled-content> images is valuable to both initial treatment planning and adaptive planning, but manual contouring is laborious and inconsistent. A novel method based on the generative adversarial network (GAN<\/jats:styled-content>) with shape constraint (SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content>) is developed for fully automated H&N OAR<\/jats:styled-content>s segmentation on CT<\/jats:styled-content> and low\u2010field MRI<\/jats:styled-content>.<\/jats:p><\/jats:sec>Methods and material<\/jats:title>A deep supervised fully convolutional DenseNet is employed as the segmentation network for voxel\u2010wise prediction. A convolutional neural network (CNN<\/jats:styled-content>)\u2010based discriminator network is then utilized to correct predicted errors and image\u2010level inconsistency between the prediction and ground truth. An additional shape representation loss between the prediction and ground truth in the latent shape space is integrated into the segmentation and adversarial loss functions to reduce false positivity and constrain the predicted shapes. The proposed segmentation method was first benchmarked on a public H&N CT<\/jats:styled-content> database including 32 patients, and then on 25 0.35T MR<\/jats:styled-content> images obtained from an MR<\/jats:styled-content>\u2010guided radiotherapy system. The OAR<\/jats:styled-content>s include brainstem, optical chiasm, larynx (MR<\/jats:styled-content> only), mandible, pharynx (MR<\/jats:styled-content> only), parotid glands (both left and right), optical nerves (both left and right), and submandibular glands (both left and right, CT<\/jats:styled-content> only). The performance of the proposed SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content> was compared with GAN<\/jats:styled-content> alone and GAN<\/jats:styled-content> with the shape constraint (SC<\/jats:styled-content>) but without the DenseNet (SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content>\u2010ResNet) to quantify the contributions of shape constraint and DenseNet in the deep neural network segmentation.<\/jats:p><\/jats:sec>Results<\/jats:title>The proposed SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content> slightly but consistently improve the segmentation accuracy on the benchmark H&N CT<\/jats:styled-content> images compared with our previous deep segmentation network, which outperformed other published methods on the same or similar CT<\/jats:styled-content> H&N dataset. On the low\u2010field MR<\/jats:styled-content> dataset, the following average Dice's indices were obtained using improved SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content>: 0.916 (brainstem), 0.589 (optical chiasm), 0.816 (mandible), 0.703 (optical nerves), 0.799 (larynx), 0.706 (pharynx), and 0.845 (parotid glands). The average surface distances ranged from 0.68\u00a0mm (brainstem) to 1.70\u00a0mm (larynx). The 95% surface distance ranged from 1.48\u00a0mm (left optical nerve) to 3.92\u00a0mm (larynx). Compared with CT<\/jats:styled-content>, using 95% surface distance evaluation, the automated segmentation accuracy is higher on MR<\/jats:styled-content> for the brainstem, optical chiasm, optical nerves and parotids, and lower for the mandible. The SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content> performance is superior to SC<\/jats:styled-content>\u2010GAN<\/jats:styled-content>\u2010ResNet, which is more accurate than GAN<\/jats:styled-content> alone on both the CT<\/jats:styled-content> and MR<\/jats:styled-content> datasets. The segmentation time for one patient is 14\u00a0seconds using a single GPU<\/jats:styled-content>.<\/jats:p><\/jats:sec>Conclusion<\/jats:title>The performance of our previous shape constrained fully CNNs for H&N segmentation is further improved by incorporating GAN<\/jats:styled-content> and DenseNet. With the novel segmentation method, we showed that the low\u2010field MR<\/jats:styled-content> images acquired on a MR<\/jats:styled-content>\u2010guided radiation radiotherapy system can support accurate and fully automated segmentation of both bony and soft tissue OAR<\/jats:styled-content>s for adaptive radiotherapy.<\/jats:p><\/jats:sec>","DOI":"10.1002\/mp.13553","type":"journal-article","created":{"date-parts":[[2019,4,19]],"date-time":"2019-04-19T09:49:16Z","timestamp":1555667356000},"page":"2669-2682","update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":60,"title":["Shape constrained fully convolutional DenseNet with adversarial training for multiorgan segmentation on head and neck CT<\/scp> and low\u2010field MR<\/scp> images"],"prefix":"10.1002","volume":"46","author":[{"given":"Nuo","family":"Tong","sequence":"first","affiliation":[{"name":"Key Lab of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi'an, Shaanxi 710071 China"},{"name":"Department of Radiation Oncology University of California\u2014Los Angeles Los Angeles CA 90095 USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shuiping","family":"Gou","sequence":"additional","affiliation":[{"name":"Key Lab of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi'an, Shaanxi 710071 China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shuyuan","family":"Yang","sequence":"additional","affiliation":[{"name":"Key Lab of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi'an, Shaanxi 710071 China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Minsong","family":"Cao","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology University of California\u2014Los Angeles Los Angeles CA 90095 USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ke","family":"Sheng","sequence":"additional","affiliation":[{"name":"Department of Radiation Oncology University of California\u2014Los Angeles Los Angeles CA 90095 USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2019,5,6]]},"reference":[{"key":"e_1_2_7_2_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.oraloncology.2012.06.019"},{"key":"e_1_2_7_3_1","doi-asserted-by":"publisher","DOI":"10.1002\/hed.24338"},{"key":"e_1_2_7_4_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijrobp.2010.10.019"},{"key":"e_1_2_7_5_1","doi-asserted-by":"publisher","DOI":"10.1080\/0284186X.2018.1505053"},{"key":"e_1_2_7_6_1","doi-asserted-by":"publisher","DOI":"10.1259\/bjr.20160667"},{"key":"e_1_2_7_7_1","doi-asserted-by":"crossref","unstructured":"HanX HoogemanMS LevendagPC et\u00a0al.Atlas\u2010based auto\u2010segmentation of head and neck CT images. Medical Image Computing and Computer\u2010Assisted Intervention.2008:434\u2013441.","DOI":"10.1007\/978-3-540-85990-1_52"},{"key":"e_1_2_7_8_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijrobp.2004.08.055"},{"key":"e_1_2_7_9_1","doi-asserted-by":"publisher","DOI":"10.1109\/34.927467"},{"key":"e_1_2_7_10_1","doi-asserted-by":"publisher","DOI":"10.1006\/cviu.1995.1004"},{"key":"e_1_2_7_11_1","doi-asserted-by":"publisher","DOI":"10.1118\/1.4871623"},{"key":"e_1_2_7_12_1","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2017.2768621"},{"key":"e_1_2_7_13_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2016.2536809"},{"key":"e_1_2_7_14_1","doi-asserted-by":"crossref","unstructured":"KawaharaJ BenTaiebA HamarnehG.Deep features to classify skin lesions. Int Symp Biomed Imaging.2016;1397\u20131400.","DOI":"10.1109\/ISBI.2016.7493528"},{"key":"e_1_2_7_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2016.2528129"},{"key":"e_1_2_7_16_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2018.2845918"},{"key":"e_1_2_7_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2018.2835303"},{"key":"e_1_2_7_18_1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2016.2548501"},{"key":"e_1_2_7_19_1","doi-asserted-by":"crossref","unstructured":"MateusD SimonovskyM NavabN KomodakisN.A deep metric for multimodal registration. Medical Image Computing and Computer\u2010Assisted Intervention \u2013 MICCAI 2016.2016;9902:10\u201318.","DOI":"10.1007\/978-3-319-46726-9_2"},{"key":"e_1_2_7_20_1","doi-asserted-by":"crossref","unstructured":"OktayO BaiW LeeM et\u00a0al.Multi\u2010input cardiac image super\u2010resolution using convolutional neural networks. Medical Image Computing and Computer\u2010Assisted Intervention \u2013 MICCAI 2016.2016:246\u2013254.","DOI":"10.1007\/978-3-319-46726-9_29"},{"key":"e_1_2_7_21_1","doi-asserted-by":"publisher","DOI":"10.1117\/12.2217162"},{"key":"e_1_2_7_22_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.12045"},{"key":"e_1_2_7_23_1","unstructured":"KayalibayB JensenG vandSP.CNN\u2010based segmentation of medical imaging data.arXiv Prepr. arXiv1701.030562017."},{"key":"e_1_2_7_24_1","first-page":"1","article-title":"Automatic segmentation of acute ischemic stroke from DWI using 3D fully convolutional DenseNets","volume":"0062","author":"Zhang R","year":"2018","journal-title":"IEEE Trans Med Imaging"},{"key":"e_1_2_7_25_1","doi-asserted-by":"publisher","DOI":"10.1002\/mp.13147"},{"key":"e_1_2_7_26_1","doi-asserted-by":"crossref","unstructured":"UrbanS TanacsA.Atlas\u2010based global and local RF segmentation of head and neck organs on multimodal MRI images. Proceedings of the 10th International Symposium on Image and Signal Processing and Analysis.2017;(Ispa):99\u2013103.","DOI":"10.1109\/ISPA.2017.8073577"},{"key":"e_1_2_7_27_1","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6560\/aacb65"},{"key":"e_1_2_7_28_1","unstructured":"GoodfellowIJ Pouget\u2010abadieJ MirzaM XuB Warde\u2010farleyD.Generative adversarial nets. Advances in Neural Information Processing Systems.2014;2672\u20132680."},{"key":"e_1_2_7_29_1","doi-asserted-by":"crossref","unstructured":"NavabN HorneggerJ WellsWM FrangiAF.U\u2010Net: convolutional networks for biomedical image segmentation. Medical Image Computing and Computer\u2010Assisted Intervention \u2013 MICCAI 2015.2015;234\u2013241.","DOI":"10.1007\/978-3-319-24574-4_28"},{"key":"e_1_2_7_30_1","doi-asserted-by":"crossref","unstructured":"AbdulkadirA LienkampSS BroxT RonnebergerO.3D U\u2010Net\u202f: learning dense volumetric segmentation from sparse annotation. Medical Image Computing and Computer\u2010Assisted Intervention \u2013 MICCAI 2016.2016;424\u2013432.","DOI":"10.1007\/978-3-319-46723-8_49"},{"key":"e_1_2_7_31_1","doi-asserted-by":"crossref","unstructured":"MilletariF.V\u2010Net\u202f: fully convolutional neural networks for volumetric medical image segmentation. 3D Vision (3DV) 2016 Fourth International Conference on IEEE.2016;567\u2013573.","DOI":"10.1109\/3DV.2016.79"},{"key":"e_1_2_7_32_1","unstructured":"GlorotX BengioY.Understanding the difficulty of training deep feedforward neural networks. International Conference Artificial Intelligence and Statistics.2010;249\u2013256."},{"key":"e_1_2_7_33_1","doi-asserted-by":"crossref","unstructured":"HeK SunJ.Deep residual learning for image recognition. IEEE Conference on Computer Vision and Pattern Recognition2016;770\u2013778.","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_2_7_34_1","doi-asserted-by":"crossref","unstructured":"HuangG Van DerML WeinbergerKQ.Densely connected convolutional networks. IEEE Conference on Computer Vision and Pattern Recognition2017;4700\u20134708.","DOI":"10.1109\/CVPR.2017.243"},{"key":"e_1_2_7_35_1","unstructured":"SzegedyC ComSG.Batch normalization\u202f: accelerating deep network training by reducing internal covariate shift. ICML.2015;448\u2013456."},{"key":"e_1_2_7_36_1","unstructured":"HintonGE.Rectified linear units improve restricted Boltzmann machines. ICML.2010;807\u2013814."},{"key":"e_1_2_7_37_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-66182-7_24"},{"key":"e_1_2_7_38_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2018.08.005"},{"key":"e_1_2_7_39_1","doi-asserted-by":"crossref","unstructured":"MahasseniB LamM TodorovicS.Unsupervised video summarization with adversarial LSTM networks. IEEE Conference on Computer Vision and Pattern Recognition.2017;202\u2013211.","DOI":"10.1109\/CVPR.2017.318"},{"key":"e_1_2_7_40_1","doi-asserted-by":"publisher","DOI":"10.1200\/JCO.2013.53.5633"},{"key":"e_1_2_7_41_1","doi-asserted-by":"publisher","DOI":"10.2307\/1932409"},{"key":"e_1_2_7_42_1","doi-asserted-by":"crossref","unstructured":"Mannion\u2010haworthR BowesM AshmanA GuillardG BrettA VincentG.Fully automatic segmentation of head and neck organs using active appearance models. MIDAS J.2016;http:\/\/www.midasjournal.org\/browse\/publication\/967.","DOI":"10.54294\/e86siq"}],"container-title":["Medical Physics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.13553","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/full-xml\/10.1002\/mp.13553","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/mp.13553","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,9,5]],"date-time":"2023-09-05T13:16:23Z","timestamp":1693919783000},"score":1,"resource":{"primary":{"URL":"https:\/\/aapm.onlinelibrary.wiley.com\/doi\/10.1002\/mp.13553"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2019,5,6]]},"references-count":41,"journal-issue":{"issue":"6","published-print":{"date-parts":[[2019,6]]}},"alternative-id":["10.1002\/mp.13553"],"URL":"https:\/\/doi.org\/10.1002\/mp.13553","archive":["Portico"],"relation":{},"ISSN":["0094-2405","2473-4209"],"issn-type":[{"value":"0094-2405","type":"print"},{"value":"2473-4209","type":"electronic"}],"subject":[],"published":{"date-parts":[[2019,5,6]]},"assertion":[{"value":"2018-11-20","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-04-15","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2019-05-06","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}