Methods for Analyzing GCaMP Calcium Imaging

GCaMP is a genetically encoded calcium indicator that is most widely used to observe the activities of populations of as many as thousands of neurons simultaneously using modern fluorescence imaging techniques. GCaMP is widely used to monitor neural activity in living animals. For example, researchers can use GCaMP to observe calcium transients in neurons, which are proxies for action potentials and synaptic activity.(GCaMP - an Overview ScienceDirect Topics, 2015). However, calcium imaging itself has several limitations. First,the calcium imaging has a poor signal to noise ratio(SNR) that makes the imaging difficult to detect dynamic signal and subtle fluctuations. Second, the fluorescence of GCaMP is slower than the action potentials from neuron activities especially in quick succession. The third issue is that there are millions of neurons that researchers investigated, but each neuron presents different activity and relation. Therefore, accurately targeting a specific group of neurons that perform similar tasks is challenging in the experiment. Aimed at these challenges,four successive methods including High pass filter, Gaussian Mixed Model,correlation matrix, and devolution were used to improve the analysis of neuron activity. By using these methods to analyze the data set from two GCaMP6 (a particular version of GCaMP) fluorescence recording data sets containing the time series traces of hundreds of neurons in the mouse primary visual cortex (Vl) residing within a three-dimensional volume approximately 800m 800m 100m in size. The results from the data analysis showed that the filtering effect of FIR high pass filter is the most significant because it significantly enhance the SNR and reduce noise. Through Gaussian Mixed Model and correlation coefficient, it clearly presents the connectivity of each neuron in a 233 times 233 matrix R and indicates the distribution of neuron activities by fitting into the Gaussian curve. The deconvolution successfully infer potential spikes. These methods efficiently enhance the noise reduction, network connectivity and temporal resolution of the analysis of imaging.