{"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/adversarial-and-perceptual-refinement-for","title":"Adversarial and Perceptual Refinement for Compressed Sensing MRI Reconstruction","arxiv_id":"1806.11216","date":"2018-06-28","proceeding":null,"authors":["Maximilian Seitzer","Guang Yang","Jo Schlemper","Ozan Oktay","Tobias Würfl","Vincent Christlein","Tom Wong","Raad Mohiaddin","David Firmin","Jennifer Keegan","Daniel Rueckert","Andreas Maier"],"abstract":"Deep learning approaches have shown promising performance for compressed\nsensing-based Magnetic Resonance Imaging. While deep neural networks trained\nwith mean squared error (MSE) loss functions can achieve high peak signal to\nnoise ratio, the reconstructed images are often blurry and lack sharp details,\nespecially for higher undersampling rates. Recently, adversarial and perceptual\nloss functions have been shown to achieve more visually appealing results.\nHowever, it remains an open question how to (1) optimally combine these loss\nfunctions with the MSE loss function and (2) evaluate such a perceptual\nenhancement. In this work, we propose a hybrid method, in which a visual\nrefinement component is learnt on top of an MSE loss-based reconstruction\nnetwork. In addition, we introduce a semantic interpretability score, measuring\nthe visibility of the region of interest in both ground truth and reconstructed\nimages, which allows us to objectively quantify the usefulness of the image\nquality for image post-processing and analysis. Applied on a large cardiac MRI\ndataset simulated with 8-fold undersampling, we demonstrate significant\nimprovements ($p<0.01$) over the state-of-the-art in both a human observer\nstudy and the semantic interpretability score.","url_abs":"http://arxiv.org/abs/1806.11216v1","url_pdf":"http://arxiv.org/pdf/1806.11216v1.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":"adversarial-and-perceptual-refinement-for","repo_url":"https://github.com/mseitzer/csmri-refinement","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"mri-reconstruction","task_name":"MRI Reconstruction"},{"task_slug":"open-question","task_name":"Open-Ended Question Answering"},{"task_slug":"compressed-sensing","task_name":"compressed sensing"}],"methods":[{"method_slug":"interpretability","method_name":"Interpretability"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.11216","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}