Spiking Sensory Neurons for Analyzing Electrophysiological Data

Low power consuming biomimetic neurons are considered for use in analyzing electrophysiological data. Starting with a circuit model of a Morris-Lecar inspired spiking neuron, we first investigate the dynamic properties. We demonstrate some of its neuro-computational features including type I and type II excitability, tonic and phasic spiking, spike latency and integration. Electroencephalogram (EEG) signals are then used as excitatory input currents and it is shown that the spiking neurons can provide new insights into brain function. The spike rates of the neurons are employed in a classification task and shown to yield similar performance compared to one using the frequency dependence. We discuss how this circuit has the potential to significantly reduce EEG data, improve privacy and lower power consumption for portable EEG systems.