{"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/deeply-supervised-rotation-equivariant","title":"Deeply Supervised Rotation Equivariant Network for Lesion Segmentation in Dermoscopy Images","arxiv_id":"1807.02804","date":"2018-07-08","proceeding":null,"authors":["Xiaomeng Li","Lequan Yu","Chi-Wing Fu","Pheng-Ann Heng"],"abstract":"Automatic lesion segmentation in dermoscopy images is an essential step for\ncomputer-aided diagnosis of melanoma. The dermoscopy images exhibits rotational\nand reflectional symmetry, however, this geometric property has not been\nencoded in the state-of-the-art convolutional neural networks based skin lesion\nsegmentation methods. In this paper, we present a deeply supervised rotation\nequivariant network for skin lesion segmentation by extending the recent group\nrotation equivariant network~\\cite{cohen2016group}. Specifically, we propose\nthe G-upsampling and G-projection operations to adapt the rotation equivariant\nclassification network for our skin lesion segmentation problem. To further\nincrease the performance, we integrate the deep supervision scheme into our\nproposed rotation equivariant segmentation architecture. The whole framework is\nequivariant to input transformations, including rotation and reflection, which\nimproves the network efficiency and thus contributes to the segmentation\nperformance. We extensively evaluate our method on the ISIC 2017 skin lesion\nchallenge dataset. The experimental results show that our rotation equivariant\nnetworks consistently excel the regular counterparts with the same model\ncomplexity under different experimental settings. Our best model achieves\n77.23\\%(JA) on the test dataset, outperforming the state-of-the-art challenging\nmethods and further demonstrating the effectiveness of our proposed deeply\nsupervised rotation equivariant segmentation network. Our best model also\noutperforms the state-of-the-art challenging methods, which further demonstrate\nthe effectiveness of our proposed deeply supervised rotation equivariant\nsegmentation network.","url_abs":"http://arxiv.org/abs/1807.02804v1","url_pdf":"http://arxiv.org/pdf/1807.02804v1.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":"deeply-supervised-rotation-equivariant","repo_url":"https://github.com/xmengli999/Deeply-Supervised-Rotation-Equivariant-Network-for-Lesion-Segmentation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"lesion-segmentation","task_name":"Lesion Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"skin-lesion-segmentation","task_name":"Skin Lesion Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}