{"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/deepdrr-a-catalyst-for-machine-learning-in","title":"DeepDRR -- A Catalyst for Machine Learning in Fluoroscopy-guided Procedures","arxiv_id":"1803.08606","date":"2018-03-22","proceeding":null,"authors":["Mathias Unberath","Jan-Nico Zaech","Sing Chun Lee","Bastian Bier","Javad Fotouhi","Mehran Armand","Nassir Navab"],"abstract":"Machine learning-based approaches outperform competing methods in most\ndisciplines relevant to diagnostic radiology. Interventional radiology,\nhowever, has not yet benefited substantially from the advent of deep learning,\nin particular because of two reasons: 1) Most images acquired during the\nprocedure are never archived and are thus not available for learning, and 2)\neven if they were available, annotations would be a severe challenge due to the\nvast amounts of data. When considering fluoroscopy-guided procedures, an\ninteresting alternative to true interventional fluoroscopy is in silico\nsimulation of the procedure from 3D diagnostic CT. In this case, labeling is\ncomparably easy and potentially readily available, yet, the appropriateness of\nresulting synthetic data is dependent on the forward model. In this work, we\npropose DeepDRR, a framework for fast and realistic simulation of fluoroscopy\nand digital radiography from CT scans, tightly integrated with the software\nplatforms native to deep learning. We use machine learning for material\ndecomposition and scatter estimation in 3D and 2D, respectively, combined with\nanalytic forward projection and noise injection to achieve the required\nperformance. On the example of anatomical landmark detection in X-ray images of\nthe pelvis, we demonstrate that machine learning models trained on DeepDRRs\ngeneralize to unseen clinically acquired data without the need for re-training\nor domain adaptation. Our results are promising and promote the establishment\nof machine learning in fluoroscopy-guided procedures.","url_abs":"http://arxiv.org/abs/1803.08606v1","url_pdf":"http://arxiv.org/pdf/1803.08606v1.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":"deepdrr-a-catalyst-for-machine-learning-in","repo_url":"https://github.com/arcadelab/DeepDRR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"deepdrr-a-catalyst-for-machine-learning-in","repo_url":"https://github.com/mathiasunberath/DeepDRR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"anatomical-landmark-detection","task_name":"Anatomical Landmark Detection"},{"task_slug":"machine-learning","task_name":"BIG-bench Machine Learning"},{"task_slug":"diagnostic","task_name":"Diagnostic"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}