Abstract:
CMOS-based microelectrode arrays (MEAs) are used to record the electrical activity in neural tissues down to micron-scale cellular structures at high spatiotemporal resolution. Continuous recording of extracellular voltages would, however generate large datasets with very sparse spatial and temporal information. Towards an efficient strategy, we propose here a Field Programmable Gate Array (FPGA) which filters the continuous CMOS MEA data stream sampled at 28 kHz and extracts electrophysiological relevant information.
In a first step, sensors of interest are selected based on the electrical label-free identification of those sensors covered by the neural tissue via adhesion noise spectroscopy. The adhesion noise-based electrical imaging is validated against light microscopic images. The FPGA finite impulse response (FIR)-filtered data is validated against software-based post-processed data.
In a second step, we implement a spike-triggered average (STA) algorithm to identify and visualize electrical activity at subcellular resolution in retinal neurons, which allows for the tracking of axonal signal propagation within the neural tissue.
This label-free, non-invasive method enables the localization of sensors of interest for electrophysiological recordings and the extraction of neuronal signals. It represents a significant advancement in neuroscience tools, which facilitates the study of neuronal network dynamics at unprecedented spatiotemporal resolution.