{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/automatic-3d-cardiovascular-mr-segmentation","title":"Automatic 3D Cardiovascular MR Segmentation with Densely-Connected Volumetric ConvNets","arxiv_id":"1708.00573","date":"2017-08-02","proceeding":null,"authors":["Lequan Yu","Jie-Zhi Cheng","Qi Dou","Xin Yang","Hao Chen","Jing Qin","Pheng-Ann Heng"],"abstract":"Automatic and accurate whole-heart and great vessel segmentation from 3D\ncardiac magnetic resonance (MR) images plays an important role in the\ncomputer-assisted diagnosis and treatment of cardiovascular disease. However,\nthis task is very challenging due to ambiguous cardiac borders and large\nanatomical variations among different subjects. In this paper, we propose a\nnovel densely-connected volumetric convolutional neural network, referred as\nDenseVoxNet, to automatically segment the cardiac and vascular structures from\n3D cardiac MR images. The DenseVoxNet adopts the 3D fully convolutional\narchitecture for effective volume-to-volume prediction. From the learning\nperspective, our DenseVoxNet has three compelling advantages. First, it\npreserves the maximum information flow between layers by a densely-connected\nmechanism and hence eases the network training. Second, it avoids learning\nredundant feature maps by encouraging feature reuse and hence requires fewer\nparameters to achieve high performance, which is essential for medical\napplications with limited training data. Third, we add auxiliary side paths to\nstrengthen the gradient propagation and stabilize the learning process. We\ndemonstrate the effectiveness of DenseVoxNet by comparing it with the\nstate-of-the-art approaches from HVSMR 2016 challenge in conjunction with\nMICCAI, and our network achieves the best dice coefficient. We also show that\nour network can achieve better performance than other 3D ConvNets but with\nfewer parameters.","url_abs":"http://arxiv.org/abs/1708.00573v1","url_pdf":"http://arxiv.org/pdf/1708.00573v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"automatic-3d-cardiovascular-mr-segmentation","repo_url":"https://github.com/yulequan/HeartSeg","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null},{"paper_slug":"automatic-3d-cardiovascular-mr-segmentation","repo_url":"https://github.com/black0017/MedicalZooPytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}