A Scalable Readout IC Based on Wideband Noise Cancelling for Full-Rate Scanning of High-Density Microelectrode Arrays

This paper presents a highly scalable readout IC for high-density microelectrode arrays (MEAs). Although the recent development of large-scale high-density MEAs provides opportunities to achieve sub-cellular neural recording over a wide network area, it is challenging to implement the readout IC that can operate with such MEAs. The requirement of high-speed recording in large-scale arrays induces wideband-noise folding, which makes it challenging to achieve a good noise performance for high-fidelity neural recording. Moreover, for the wideband readout, the major noise contributor changes from the readout circuit to the cell-electrode interface. In this paper, we first show why the interface noise becomes the dominant noise source and elucidate its component that contributes the most: sealing resistance. Then, we propose a new readout circuit structure, which can effectively cancel the wideband interface noise. As a result, the signal-to-noise ratio of input neural spike signals is improved dramatically in all cell-attachment or sealing conditions. Particularly, it is shown that under weakly sealed conditions, the spikes can be detected only when the proposed wideband noise cancellation technique is applied.