{"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/improving-automated-multiple-sclerosis-lesion","title":"Improving automated multiple sclerosis lesion segmentation with a cascaded 3D convolutional neural network approach","arxiv_id":"1702.04869","date":"2017-02-16","proceeding":null,"authors":["Sergi Valverde","Mariano Cabezas","Eloy Roura","Sandra González-Villà","Deborah Pareto","Joan-Carles Vilanova","Lluís Ramió-Torrentà","Àlex Rovira","Arnau Oliver","Xavier Lladó"],"abstract":"In this paper, we present a novel automated method for White Matter (WM)\nlesion segmentation of Multiple Sclerosis (MS) patient images. Our approach is\nbased on a cascade of two 3D patch-wise convolutional neural networks (CNN).\nThe first network is trained to be more sensitive revealing possible candidate\nlesion voxels while the second network is trained to reduce the number of\nmisclassified voxels coming from the first network. This cascaded CNN\narchitecture tends to learn well from small sets of training data, which can be\nvery interesting in practice, given the difficulty to obtain manual label\nannotations and the large amount of available unlabeled Magnetic Resonance\nImaging (MRI) data. We evaluate the accuracy of the proposed method on the\npublic MS lesion segmentation challenge MICCAI2008 dataset, comparing it with\nrespect to other state-of-the-art MS lesion segmentation tools. Furthermore,\nthe proposed method is also evaluated on two private MS clinical datasets,\nwhere the performance of our method is also compared with different recent\npublic available state-of-the-art MS lesion segmentation methods. At the time\nof writing this paper, our method is the best ranked approach on the MICCAI2008\nchallenge, outperforming the rest of 60 participant methods when using all the\navailable input modalities (T1-w, T2-w and FLAIR), while still in the top-rank\n(3rd position) when using only T1-w and FLAIR modalities. On clinical MS data,\nour approach exhibits a significant increase in the accuracy segmenting of WM\nlesions when compared with the rest of evaluated methods, highly correlating\n($r \\ge 0.97$) also with the expected lesion volume.","url_abs":"http://arxiv.org/abs/1702.04869v1","url_pdf":"http://arxiv.org/pdf/1702.04869v1.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":"improving-automated-multiple-sclerosis-lesion","repo_url":"https://github.com/NIC-VICOROB/cnn-ms-lesion-segmentation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"improving-automated-multiple-sclerosis-lesion","repo_url":"https://github.com/sergivalverde/cnn-ms-lesion-segmentation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"lesion-segmentation","task_name":"Lesion Segmentation"},{"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}