{"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/an-application-of-cascaded-3d-fully","title":"An application of cascaded 3D fully convolutional networks for medical image segmentation","arxiv_id":"1803.05431","date":"2018-03-14","proceeding":null,"authors":["Holger R. Roth","Hirohisa ODA","Xiangrong Zhou","Natsuki Shimizu","Ying Yang","Yuichiro Hayashi","Masahiro Oda","Michitaka Fujiwara","Kazunari Misawa","Kensaku MORI"],"abstract":"Recent advances in 3D fully convolutional networks (FCN) have made it\nfeasible to produce dense voxel-wise predictions of volumetric images. In this\nwork, we show that a multi-class 3D FCN trained on manually labeled CT scans of\nseveral anatomical structures (ranging from the large organs to thin vessels)\ncan achieve competitive segmentation results, while avoiding the need for\nhandcrafting features or training class-specific models.\n  To this end, we propose a two-stage, coarse-to-fine approach that will first\nuse a 3D FCN to roughly define a candidate region, which will then be used as\ninput to a second 3D FCN. This reduces the number of voxels the second FCN has\nto classify to ~10% and allows it to focus on more detailed segmentation of the\norgans and vessels.\n  We utilize training and validation sets consisting of 331 clinical CT images\nand test our models on a completely unseen data collection acquired at a\ndifferent hospital that includes 150 CT scans, targeting three anatomical\norgans (liver, spleen, and pancreas). In challenging organs such as the\npancreas, our cascaded approach improves the mean Dice score from 68.5 to\n82.2%, achieving the highest reported average score on this dataset. We compare\nwith a 2D FCN method on a separate dataset of 240 CT scans with 18 classes and\nachieve a significantly higher performance in small organs and vessels.\nFurthermore, we explore fine-tuning our models to different datasets.\n  Our experiments illustrate the promise and robustness of current 3D FCN based\nsemantic segmentation of medical images, achieving state-of-the-art results.\nOur code and trained models are available for download:\nhttps://github.com/holgerroth/3Dunet_abdomen_cascade.","url_abs":"http://arxiv.org/abs/1803.05431v2","url_pdf":"http://arxiv.org/pdf/1803.05431v2.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":"an-application-of-cascaded-3d-fully","repo_url":"https://github.com/holgerroth/3Dunet_abdomen_cascade","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"3d-medical-imaging-segmentation","task_name":"3D Medical Imaging Segmentation"},{"task_slug":"image-segmentation","task_name":"Image Segmentation"},{"task_slug":"medical-image-segmentation","task_name":"Medical Image Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"fcn","method_name":"FCN"},{"method_slug":"max-pooling","method_name":"Max Pooling"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-medical-imaging-segmentation-on-tcia","task":"3D Medical Imaging Segmentation","dataset":"TCIA Pancreas-CT","model":"Multi-class 3D FCN","rank_in_archive_order":2,"of":2,"metrics":{"Dice Score":"76.8"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}