Noise Characterization of Electrophysiological Signals Recorded from High Resolution Brain-Chip Interface

The evoked potentials recorded using high resolution chips from a brain-chip interface provide very useful information in understanding the brain activity with greater details. In many cases, to maximize the signal detection, high precision electronics devices are employed near the recording site. This causes the recordings to be contaminated with unavoidable electrical or thermal noises. To assess the quality of the recordings in terms of signal-to-noise ratio (SNR), an automated noise characterization system is required. This paper presents a novel, automated and suitable method for noise characterization of the neuronal recordings from brain-chip interface. The method first detects the first portion of the signal without any evoked neuronal activity (steady-state) and then uses the classical measurement error model to calculate the measurement error present in the steady-state. It also provides other important statistical information about the signal, like - the mean, standard deviation and the statistical distribution of the detected steady-state, i.e., the noise. Above all, its features like - steady-state detection, easy implementation, and computational efficiency makes it suitable for neuronal recordings using high resolution brain-chip interface.