{"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/spatially-localized-atlas-network-tiles","title":"Spatially Localized Atlas Network Tiles Enables 3D Whole Brain Segmentation from Limited Data","arxiv_id":"1806.00546","date":"2018-06-01","proceeding":null,"authors":["Yuankai Huo","Zhoubing Xu","Katherine Aboud","Prasanna Parvathaneni","Shunxing Bao","Camilo Bermudez","Susan M. Resnick","Laurie E. Cutting","Bennett A. Landman"],"abstract":"Whole brain segmentation on a structural magnetic resonance imaging (MRI) is\nessential in non-invasive investigation for neuroanatomy. Historically,\nmulti-atlas segmentation (MAS) has been regarded as the de facto standard\nmethod for whole brain segmentation. Recently, deep neural network approaches\nhave been applied to whole brain segmentation by learning random patches or 2D\nslices. Yet, few previous efforts have been made on detailed whole brain\nsegmentation using 3D networks due to the following challenges: (1) fitting\nentire whole brain volume into 3D networks is restricted by the current GPU\nmemory, and (2) the large number of targeting labels (e.g., > 100 labels) with\nlimited number of training 3D volumes (e.g., < 50 scans). In this paper, we\npropose the spatially localized atlas network tiles (SLANT) method to\ndistribute multiple independent 3D fully convolutional networks to cover\noverlapped sub-spaces in a standard atlas space. This strategy simplifies the\nwhole brain learning task to localized sub-tasks, which was enabled by combing\ncanonical registration and label fusion techniques with deep learning. To\naddress the second challenge, auxiliary labels on 5111 initially unlabeled\nscans were created by MAS for pre-training. From empirical validation, the\nstate-of-the-art MAS method achieved mean Dice value of 0.76, 0.71, and 0.68,\nwhile the proposed method achieved 0.78, 0.73, and 0.71 on three validation\ncohorts. Moreover, the computational time reduced from > 30 hours using MAS to\n~15 minutes using the proposed method. The source code is available online\nhttps://github.com/MASILab/SLANTbrainSeg","url_abs":"http://arxiv.org/abs/1806.00546v2","url_pdf":"http://arxiv.org/pdf/1806.00546v2.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":"spatially-localized-atlas-network-tiles","repo_url":"https://github.com/MASILab/SLANTbrainSeg","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"caffe2","reach":null},{"paper_slug":"spatially-localized-atlas-network-tiles","repo_url":"https://github.com/MASILab/SLANT_brain_seg","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":null}],"tasks":[{"task_slug":"brain-segmentation","task_name":"Brain Segmentation"},{"task_slug":null,"task_name":"GPU"},{"task_slug":"segmentation","task_name":"Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}