Inferring Neuronal Dynamics from Calcium Imaging Data Using Biophysical Models and Bayesian Inference

• DOI: 10.1371/journal.pcbi.1004736
• Publication Date: 2016-02-19T18:49:53Z
• Authors: Vahid Rahmati; Knut Kirmse; Dimitrije Marković; Knut Holthoff; Stefan J. Kiebel
• Keywords: Bursting; Calcium imaging; TRACE (psycholinguistics); Temporal resolution

Calcium imaging has been used as a promising technique to monitor the dynamic activity of neuronal populations. However, calcium trace is temporally smeared which restricts extraction quantities interest such spike trains individual neurons. To address this issue, reconstruction algorithms have introduced. One limitation reconstructions that underlying models are not informed about biophysics and burst generations. Such existing prior knowledge might be useful for constraining possible solutions spikes. Here we describe, in novel Bayesian approach, how principled dynamics can employed infer biophysical variables parameters from fluorescence traces. By using both synthetic vitro recorded traces, demonstrate new approach able reconstruct different repetitive spiking and/or bursting patterns with accurate single resolution. Furthermore, show high inference precision preserved even if rather noisy or transients slow rise kinetics lasting several hundred milliseconds, inhomogeneous decay times. In addition, discuss use inferring parameter changes, e.g. due pharmacological intervention, well complex characteristics immature circuits.