Constrained matrix factorization methods for denoising and demixing voltage imaging data

Voltage imaging (VI) is an emerging technique for recording neural population activity that offers advantages over tried-and-true calcium imaging (CI). Namely, voltage indicator dynamics (1) allow for higher temporal resolution imaging and (2) are sensitive to sub-threshold voltage fluctuations which calcium indicators cannot capture. We develop a new data processing pipeline specialized for VI data and demonstrate the ability to resolve single cell activity to a high degree of spatial and temporal acuity on a challenging in vitro dataset with highly synchronous activity driven by optogenetic stimulation.