Miniscope Processing Suite (MPS): An Intuitive, No-Code, Scalable Pipeline for Long Duration Calcium Imaging

Abstract

Miniscope calcium imaging provides a unique window into the activity of neurons during behavior while enabling spatial localization of individual cells across time. Despite its potential to revolutionize in vivo imaging alongside the rise of optogenetic tools, miniscopes remain underutilized. This gap may stem from the lack of an easy-to-use preprocessing software tailored to miniscope data. To address this need, we developed the first no-code end-to-end scalable pipeline for preprocessing large miniscope recordings: the Miniscope Processing Suite (MPS). MPS is the first implementation of Constrained Non-negative Matrix Factorization (CNMF) with Nonnegative Double Singular Value Decomposition (NNDSVD) initialization, multi-lasso segmentation, and parallelized temporal and spatial updates, enabling analysis of recordings exceeding three hours on a standard hardware. We tested a large dataset consisting of 28 operant behavior sessions (77 hrs, 7.26 TB of video data) on a single workstation. MPS completed the analysis in 55.6 hrs, (averaging 0.72 minute of processing time per minute of recording), which is 10-20X faster than traditional pipelines. Packaged as an easy downloadable plug-and-play software with a graphical user interface and requiring no coding or Git experience, MPS lowers the barrier to miniscope analysis while improving on current methodologies.