{"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/learning-a-probabilistic-model-for","title":"Learning a Probabilistic Model for Diffeomorphic Registration","arxiv_id":"1812.07460","date":"2018-12-18","proceeding":null,"authors":["Julian Krebs","Hervé Delingette","Boris Mailhé","Nicholas Ayache","Tommaso Mansi"],"abstract":"We propose to learn a low-dimensional probabilistic deformation model from\ndata which can be used for registration and the analysis of deformations. The\nlatent variable model maps similar deformations close to each other in an\nencoding space. It enables to compare deformations, generate normal or\npathological deformations for any new image or to transport deformations from\none image pair to any other image. Our unsupervised method is based on\nvariational inference. In particular, we use a conditional variational\nautoencoder (CVAE) network and constrain transformations to be symmetric and\ndiffeomorphic by applying a differentiable exponentiation layer with a\nsymmetric loss function. We also present a formulation that includes spatial\nregularization such as diffusion-based filters. Additionally, our framework\nprovides multi-scale velocity field estimations. We evaluated our method on 3-D\nintra-subject registration using 334 cardiac cine-MRIs. On this dataset, our\nmethod showed state-of-the-art performance with a mean DICE score of 81.2% and\na mean Hausdorff distance of 7.3mm using 32 latent dimensions compared to three\nstate-of-the-art methods while also demonstrating more regular deformation\nfields. The average time per registration was 0.32s. Besides, we visualized the\nlearned latent space and show that the encoded deformations can be used to\ntransport deformations and to cluster diseases with a classification accuracy\nof 83% after applying a linear projection.","url_abs":"http://arxiv.org/abs/1812.07460v2","url_pdf":"http://arxiv.org/pdf/1812.07460v2.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":[],"tasks":[{"task_slug":"deformable-medical-image-registration","task_name":"Deformable Medical Image Registration"},{"task_slug":"diffeomorphic-medical-image-registration","task_name":"Diffeomorphic Medical Image Registration"},{"task_slug":"medical-image-registration","task_name":"Medical Image Registration"},{"task_slug":"variational-inference","task_name":"Variational Inference"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/diffeomorphic-medical-image-registration-on-1","task":"Diffeomorphic Medical Image Registration","dataset":"Automatic Cardiac Diagnosis Challenge (ACDC)","model":"cVAE Diffeomorphic (S3)","rank_in_archive_order":1,"of":3,"metrics":{"Dice":"0.812","Grad Det-Jac":"1.4","Hausdorff Distance (mm)":"7.3","RMSE":"0.30"},"uses_additional_data":true}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1812.07460","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}