Inferring Neuronal Dynamics from Calcium Imaging Data Using Biophysical Models and Bayesian Inference
Fig 7
Results of the proposed approach for in vitro fluorescence traces with slowly rising transients (around 400 ms).
(A): QGIF model, (B): FHN model with high SNR, (C) FHN model with low SNR. First row (A-C): The in vitro fluorescence traces containing transients with relatively variable and slow rise times, mediated by spontaneous GDPs. The low SNR trace in C was generated by contaminating the in vitro trace by background noise. Second row (A-C): Inferred non-saturating [Ca2+] kinetics, Third row (A-C): Inferred membrane potentials, where the onset times of GDP-mediated events (single spike or burst) determined by electrophysiological recordings (grey stars) and inferred spikes are highly concurrent. The two light blue stars indicate that there was no recorded transmembrane current available for the observed events in the fluorescence trace(s). Note that for a GDP-mediated burst event the onset time refers to the occurrence of its first spike. Fourth row (A-C): Zoom into the slow rise time of the first fluorescence transient in the traces shown in the first row. The numbers and stars indicate the veridical spike count and the onset time of each GDP. The rise time from GDP onsets to the fluorescence transient peaks was around 300–450 ms. The decay kinetics of the transients was lasting around 3.5–4.5 seconds.