{"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/mri-tumor-segmentation-with-densely-connected","title":"MRI Tumor Segmentation with Densely Connected 3D CNN","arxiv_id":"1802.02427","date":"2018-01-18","proceeding":null,"authors":["Lele Chen","Yue Wu","Adora M. DSouza","Anas Z. Abidin","Axel Wismuller","Chenliang Xu"],"abstract":"Glioma is one of the most common and aggressive types of primary brain\ntumors. The accurate segmentation of subcortical brain structures is crucial to\nthe study of gliomas in that it helps the monitoring of the progression of\ngliomas and aids the evaluation of treatment outcomes. However, the large\namount of required human labor makes it difficult to obtain the manually\nsegmented Magnetic Resonance Imaging (MRI) data, limiting the use of precise\nquantitative measurements in the clinical practice. In this work, we try to\naddress this problem by developing a 3D Convolutional Neural Network~(3D CNN)\nbased model to automatically segment gliomas. The major difficulty of our\nsegmentation model comes with the fact that the location, structure, and shape\nof gliomas vary significantly among different patients. In order to accurately\nclassify each voxel, our model captures multi-scale contextual information by\nextracting features from two scales of receptive fields. To fully exploit the\ntumor structure, we propose a novel architecture that hierarchically segments\ndifferent lesion regions of the necrotic and non-enhancing tumor~(NCR/NET),\nperitumoral edema~(ED) and GD-enhancing tumor~(ET). Additionally, we utilize\ndensely connected convolutional blocks to further boost the performance. We\ntrain our model with a patch-wise training schema to mitigate the class\nimbalance problem. The proposed method is validated on the BraTS 2017 dataset\nand it achieves Dice scores of 0.72, 0.83 and 0.81 for the complete tumor,\ntumor core and enhancing tumor, respectively. These results are comparable to\nthe reported state-of-the-art results, and our method is better than existing\n3D-based methods in terms of compactness, time and space efficiency.","url_abs":"http://arxiv.org/abs/1802.02427v2","url_pdf":"http://arxiv.org/pdf/1802.02427v2.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":"mri-tumor-segmentation-with-densely-connected","repo_url":"https://github.com/lelechen63/MRI-tumor-segmentation-Brats","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"mri-tumor-segmentation-with-densely-connected","repo_url":"https://github.com/nesonn/3D-cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"mri-tumor-segmentation-with-densely-connected","repo_url":"https://github.com/littleMMJ/mindsporemodels","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"tumor-segmentation","task_name":"Tumor 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}