{"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/quicksilver-fast-predictive-image","title":"Quicksilver: Fast Predictive Image Registration - a Deep Learning Approach","arxiv_id":"1703.10908","date":"2017-03-31","proceeding":null,"authors":["Xiao Yang","Roland Kwitt","Martin Styner","Marc Niethammer"],"abstract":"This paper introduces Quicksilver, a fast deformable image registration\nmethod. Quicksilver registration for image-pairs works by patch-wise prediction\nof a deformation model based directly on image appearance. A deep\nencoder-decoder network is used as the prediction model. While the prediction\nstrategy is general, we focus on predictions for the Large Deformation\nDiffeomorphic Metric Mapping (LDDMM) model. Specifically, we predict the\nmomentum-parameterization of LDDMM, which facilitates a patch-wise prediction\nstrategy while maintaining the theoretical properties of LDDMM, such as\nguaranteed diffeomorphic mappings for sufficiently strong regularization. We\nalso provide a probabilistic version of our prediction network which can be\nsampled during the testing time to calculate uncertainties in the predicted\ndeformations. Finally, we introduce a new correction network which greatly\nincreases the prediction accuracy of an already existing prediction network. We\nshow experimental results for uni-modal atlas-to-image as well as uni- / multi-\nmodal image-to-image registrations. These experiments demonstrate that our\nmethod accurately predicts registrations obtained by numerical optimization, is\nvery fast, achieves state-of-the-art registration results on four standard\nvalidation datasets, and can jointly learn an image similarity measure.\nQuicksilver is freely available as an open-source software.","url_abs":"http://arxiv.org/abs/1703.10908v4","url_pdf":"http://arxiv.org/pdf/1703.10908v4.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":"quicksilver-fast-predictive-image","repo_url":"https://github.com/rkwitt/quicksilver","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"image-registration","task_name":"Image Registration"},{"task_slug":"prediction","task_name":"Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.10908","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}