{"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/a-lifelong-learning-approach-to-brain-mr","title":"A Lifelong Learning Approach to Brain MR Segmentation Across Scanners and Protocols","arxiv_id":"1805.10170","date":"2018-05-25","proceeding":null,"authors":["Neerav Karani","Krishna Chaitanya","Christian Baumgartner","Ender Konukoglu"],"abstract":"Convolutional neural networks (CNNs) have shown promising results on several\nsegmentation tasks in magnetic resonance (MR) images. However, the accuracy of\nCNNs may degrade severely when segmenting images acquired with different\nscanners and/or protocols as compared to the training data, thus limiting their\npractical utility. We address this shortcoming in a lifelong multi-domain\nlearning setting by treating images acquired with different scanners or\nprotocols as samples from different, but related domains. Our solution is a\nsingle CNN with shared convolutional filters and domain-specific batch\nnormalization layers, which can be tuned to new domains with only a few\n($\\approx$ 4) labelled images. Importantly, this is achieved while retaining\nperformance on the older domains whose training data may no longer be\navailable. We evaluate the method for brain structure segmentation in MR\nimages. Results demonstrate that the proposed method largely closes the gap to\nthe benchmark, which is training a dedicated CNN for each scanner.","url_abs":"http://arxiv.org/abs/1805.10170v1","url_pdf":"http://arxiv.org/pdf/1805.10170v1.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":"a-lifelong-learning-approach-to-brain-mr","repo_url":"https://github.com/liuquande/SAML","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"lifelong-learning","task_name":"Lifelong learning"},{"task_slug":"segmentation","task_name":"Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1805.10170","atlas_url":"https://app.syntology.ai/?focus=1805.10170","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}