{"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/anatomynet-deep-learning-for-fast-and-fully","title":"AnatomyNet: Deep Learning for Fast and Fully Automated Whole-volume Segmentation of Head and Neck Anatomy","arxiv_id":"1808.05238","date":"2018-08-15","proceeding":null,"authors":["Wentao Zhu","Yufang Huang","Liang Zeng","Xuming Chen","Yong liu","Zhen Qian","Nan Du","Wei Fan","Xiaohui Xie"],"abstract":"Methods: Our deep learning model, called AnatomyNet, segments OARs from head\nand neck CT images in an end-to-end fashion, receiving whole-volume HaN CT\nimages as input and generating masks of all OARs of interest in one shot.\nAnatomyNet is built upon the popular 3D U-net architecture, but extends it in\nthree important ways: 1) a new encoding scheme to allow auto-segmentation on\nwhole-volume CT images instead of local patches or subsets of slices, 2)\nincorporating 3D squeeze-and-excitation residual blocks in encoding layers for\nbetter feature representation, and 3) a new loss function combining Dice scores\nand focal loss to facilitate the training of the neural model. These features\nare designed to address two main challenges in deep-learning-based HaN\nsegmentation: a) segmenting small anatomies (i.e., optic chiasm and optic\nnerves) occupying only a few slices, and b) training with inconsistent data\nannotations with missing ground truth for some anatomical structures.\n  Results: We collected 261 HaN CT images to train AnatomyNet, and used MICCAI\nHead and Neck Auto Segmentation Challenge 2015 as a benchmark dataset to\nevaluate the performance of AnatomyNet. The objective is to segment nine\nanatomies: brain stem, chiasm, mandible, optic nerve left, optic nerve right,\nparotid gland left, parotid gland right, submandibular gland left, and\nsubmandibular gland right. Compared to previous state-of-the-art results from\nthe MICCAI 2015 competition, AnatomyNet increases Dice similarity coefficient\nby 3.3% on average. AnatomyNet takes about 0.12 seconds to fully segment a head\nand neck CT image of dimension 178 x 302 x 225, significantly faster than\nprevious methods. In addition, the model is able to process whole-volume CT\nimages and delineate all OARs in one pass, requiring little pre- or\npost-processing.\nhttps://github.com/wentaozhu/AnatomyNet-for-anatomical-segmentation.git.","url_abs":"http://arxiv.org/abs/1808.05238v2","url_pdf":"http://arxiv.org/pdf/1808.05238v2.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":"anatomynet-deep-learning-for-fast-and-fully","repo_url":"https://github.com/wentaozhu/AnatomyNet-for-anatomical-segmentation","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null},{"paper_slug":"anatomynet-deep-learning-for-fast-and-fully","repo_url":"https://github.com/BioWar/Satellite-Image-Segmentation-using-Deep-Learning-for-Deforestation-Detection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"3d-medical-imaging-segmentation","task_name":"3D Medical Imaging Segmentation"},{"task_slug":"anatomy","task_name":"Anatomy"},{"task_slug":"medical-image-segmentation","task_name":"Medical Image Segmentation"}],"methods":[{"method_slug":"concatenated-skip-connection","method_name":"Concatenated Skip Connection"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"focal-loss","method_name":"Focal Loss"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"u-net","method_name":"U-Net"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/medical-image-segmentation-on-miccai-2015-2","task":"Medical Image Segmentation","dataset":"MICCAI 2015 Head and Neck Challenge","model":"AnatomyNet","rank_in_archive_order":1,"of":1,"metrics":{"Dice":"79.25"},"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}