Simultaneous Denoising, Deconvolution, and Demixing of Calcium Imaging Data

Abstract

SUMMARY We present a modular approach for analyzing calcium imaging recordings of large neuronal ensembles. Our goal is to simultaneously identify the locations of the neurons, demix spatially overlapping components, and denoise and deconvolve the spiking activity from the slow dynamics of the calcium indicator. Our approach relies on a constrained nonnegative matrix factorization that expresses the spatiotemporal fluorescence activity as the product of a spatial matrix that encodes the spatial footprint of each neuron in the optical field and a temporal matrix that characterizes the calcium concentration of each neuron over time. This framework is combined with a novel constrained deconvolution approach that extracts estimates of neural activity from fluorescence traces, to create a spatiotemporal processing algorithm that requires minimal parameter tuning. We demonstrate the general applicability of our method by applying it to in vitro and in vivo multineuronal imaging data, whole-brain light-sheet imaging data, and dendritic imaging data.

ABLE: an Activity-Based Level Set Segmentation Algorithm for Two-Photon Calcium Imaging Data

“…We implicitly evolve an active contour via the level set function. The evolution of the level set 13 function is driven by a local model of the imaging data temporal activity. The data model includes no 14 assumptions on a cell's morphology or stereotypical temporal activity.…”

“…When we have an image not a video (i.e. I(x) and f are one-dimensional) this dissimilarity met- 13 ric reduces to the fitting term introduced by Chan and Vese (2001). For datasets in which the fluorescence 14 expression level varies significantly throughout cells and, as a consequence, pixels in the same cell exhibit 15 the same pattern of activity at different magnitudes, we use a measure based on the correlation, such that 16 (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.…”

“…http://dx.doi.org/10.1101/190348 doi: bioRxiv preprint first posted online Sep. 18, 2017; In particular, we assume that a pixel in multiple cells would have a time course well fit by the sum of the 1 interior time courses for each cell. The external energy in the case of multiple cells is thus by a function φ i , where φ i is positive for all pixels in the cell interior, negative for those in the narrowband 13 and zero for all pixels on the boundary (see Fig. 1C).…”

“…To attain segmentation results on the real datasets presented in this paper, 12 we use λ = 150 (Section 3.1), λ = 50 (Section 3.3), λ = 25 (Section 3.4) and λ = 10 (Section 3.5). 13 …”

Imaging, manipulation and optogenetics in zebrafish

“…Past studies have approached separately those studies [141][142][143][144]. Very few methods cover these three problems but, Pnevmatikakis et al [145] is such a method and has the advantages of dealing with big datasets by running computations in the cloud, giving clean results by discarding artefacts. It also is one of the top performers in the neurofinder competition.…”