{"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/brain-tumor-segmentation-based-on-refined","title":"Brain Tumor Segmentation Based on Refined Fully Convolutional Neural Networks with A Hierarchical Dice Loss","arxiv_id":"1712.09093","date":"2017-12-25","proceeding":null,"authors":["Jiachi Zhang","Xiaolei Shen","Tianqi Zhuo","Hong Zhou"],"abstract":"As a basic task in computer vision, semantic segmentation can provide\nfundamental information for object detection and instance segmentation to help\nthe artificial intelligence better understand real world. Since the proposal of\nfully convolutional neural network (FCNN), it has been widely used in semantic\nsegmentation because of its high accuracy of pixel-wise classification as well\nas high precision of localization. In this paper, we apply several famous FCNN\nto brain tumor segmentation, making comparisons and adjusting network\narchitectures to achieve better performance measured by metrics such as\nprecision, recall, mean of intersection of union (mIoU) and dice score\ncoefficient (DSC). The adjustments to the classic FCNN include adding more\nconnections between convolutional layers, enlarging decoders after up sample\nlayers and changing the way shallower layers' information is reused. Besides\nthe structure modification, we also propose a new classifier with a\nhierarchical dice loss. Inspired by the containing relationship between\nclasses, the loss function converts multiple classification to multiple binary\nclassification in order to counteract the negative effect caused by imbalance\ndata set. Massive experiments have been done on the training set and testing\nset in order to assess our refined fully convolutional neural networks and new\ntypes of loss function. Competitive figures prove they are more effective than\ntheir predecessors.","url_abs":"http://arxiv.org/abs/1712.09093v3","url_pdf":"http://arxiv.org/pdf/1712.09093v3.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":"brain-tumor-segmentation-based-on-refined","repo_url":"https://github.com/milliondegree/semantic-segmentation-tensorflow","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"binary-classification","task_name":"Binary Classification"},{"task_slug":"brain-tumor-segmentation","task_name":"Brain Tumor Segmentation"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"tumor-segmentation","task_name":"Tumor Segmentation"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}