{"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/penalized-matrix-decomposition-for-denoising","title":"Penalized matrix decomposition for denoising, compression, and improved demixing of functional imaging data","arxiv_id":"1807.06203","date":"2018-07-17","proceeding":null,"authors":["E. Kelly Buchanan","Ian Kinsella","Ding Zhou","Rong Zhu","Pengcheng Zhou","Felipe Gerhard","John Ferrante","Ying Ma","Sharon Kim","Mohammed Shaik","Yajie Liang","Rongwen Lu","Jacob Reimer","Paul Fahey","Taliah Muhammad","Graham Dempsey","Elizabeth Hillman","Na Ji","Andreas Tolias","Liam Paninski"],"abstract":"Calcium imaging has revolutionized systems neuroscience, providing the\nability to image large neural populations with single-cell resolution. The\nresulting datasets are quite large, which has presented a barrier to routine\nopen sharing of this data, slowing progress in reproducible research. State of\nthe art methods for analyzing this data are based on non-negative matrix\nfactorization (NMF); these approaches solve a non-convex optimization problem,\nand are effective when good initializations are available, but can break down\nin low-SNR settings where common initialization approaches fail. Here we\nintroduce an approach to compressing and denoising functional imaging data. The\nmethod is based on a spatially-localized penalized matrix decomposition (PMD)\nof the data to separate (low-dimensional) signal from (temporally-uncorrelated)\nnoise. This approach can be applied in parallel on local spatial patches and is\ntherefore highly scalable, does not impose non-negativity constraints or\nrequire stringent identifiability assumptions (leading to significantly more\nrobust results compared to NMF), and estimates all parameters directly from the\ndata, so no hand-tuning is required. We have applied the method to a wide range\nof functional imaging data (including one-photon, two-photon, three-photon,\nwidefield, somatic, axonal, dendritic, calcium, and voltage imaging datasets):\nin all cases, we observe ~2-4x increases in SNR and compression rates of\n20-300x with minimal visible loss of signal, with no adjustment of\nhyperparameters; this in turn facilitates the process of demixing the observed\nactivity into contributions from individual neurons. We focus on two\nchallenging applications: dendritic calcium imaging data and voltage imaging\ndata in the context of optogenetic stimulation. In both cases, we show that our\nnew approach leads to faster and much more robust extraction of activity from\nthe data.","url_abs":"http://arxiv.org/abs/1807.06203v1","url_pdf":"http://arxiv.org/pdf/1807.06203v1.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":"penalized-matrix-decomposition-for-denoising","repo_url":"https://github.com/paninski-lab/funimag","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}