{"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/multi-scale-gradual-integration-cnn-for-false","title":"Multi-Scale Gradual Integration CNN for False Positive Reduction in Pulmonary Nodule Detection","arxiv_id":"1807.10581","date":"2018-07-24","proceeding":null,"authors":["Bum-Chae Kim","Jun-Sik Choi","Heung-Il Suk"],"abstract":"Lung cancer is a global and dangerous disease, and its early detection is\ncrucial to reducing the risks of mortality. In this regard, it has been of\ngreat interest in developing a computer-aided system for pulmonary nodules\ndetection as early as possible on thoracic CT scans. In general, a nodule\ndetection system involves two steps: (i) candidate nodule detection at a high\nsensitivity, which captures many false positives and (ii) false positive\nreduction from candidates. However, due to the high variation of nodule\nmorphological characteristics and the possibility of mistaking them for\nneighboring organs, candidate nodule detection remains a challenge. In this\nstudy, we propose a novel Multi-scale Gradual Integration Convolutional Neural\nNetwork (MGI-CNN), designed with three main strategies: (1) to use multi-scale\ninputs with different levels of contextual information, (2) to use abstract\ninformation inherent in different input scales with gradual integration, and\n(3) to learn multi-stream feature integration in an end-to-end manner. To\nverify the efficacy of the proposed network, we conducted exhaustive\nexperiments on the LUNA16 challenge datasets by comparing the performance of\nthe proposed method with state-of-the-art methods in the literature. On two\ncandidate subsets of the LUNA16 dataset, i.e., V1 and V2, our method achieved\nan average CPM of 0.908 (V1) and 0.942 (V2), outperforming comparable methods\nby a large margin. Our MGI-CNN is implemented in Python using TensorFlow and\nthe source code is available from 'https://github.com/ku-milab/MGICNN.'","url_abs":"http://arxiv.org/abs/1807.10581v1","url_pdf":"http://arxiv.org/pdf/1807.10581v1.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":"multi-scale-gradual-integration-cnn-for-false","repo_url":"https://github.com/ku-milab/MGICNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}