Blind deconvolution for spike inference from fluorescence recordings.

J. Tubiana, , S. Wolf, , T. Panier , G. Debrégeas

Bibtex , URL
Journal of Neuroscience Methods, 342, 108763
Published 01 Aug. 2020
DOI: 10.1016/j.jneumeth.2020.108763


The parallel developments of genetically-encoded calcium indicators and fast fluorescence imaging techniques allows to simultaneously record neural activity of extended neuronal populations in vivo, opening a new arena for systems neuroscience. To fully harness the potential of functional imaging, one needs to infer the sequence of action potentials from fluorescence time traces. Here we build on recently proposed computational approaches to develop a blind sparse deconvolution algorithm (BSD) for inferring spike trains from fluorescence traces that is based on a generative model of fluorescence traces. The parameters of the generative model, such as spike amplitude, noise level and the rise and decay time constants are iteratively estimated from the fluorescence trace whereas the sparse penalty and threshold for binarizing spikes are selected to maximize expected precisionrecall performance. The algorithm supports super-resolution, and we provide theoretical bounds on the performance of the algorithm in terms of precision-recall and temporal accuracy. We establish the performance of the algorithm on synthetic data and real functional imaging data for which ground truth spikes are available. We show that the inferred spike trains and model parameters correlate very well with their ground truth counterparts, and that superresolution can signi cantly improve the temporal accuracy of the inferred spikes. Our method outperforms classical sparse deconvolution algorithms in terms of robustness, speed and/or accuracy. Overall, BSD is a fast and reliable preprocessing tool for quantitative modelling of functional imaging.

Cette publication est associée à :

Imagerie calcique et comportement du poisson zèbre