A ΔΣ Neural Frontend with Stochastic Signal Processing for Stimulation Artifact Suppression

This paper presents a novel recording front-end with integrated electrical stimulation artifact suppression using adaptive stochastic filtering. The architecture of the recorder is based on a low-noise programmable gain amplifier and a $1^{\text {st}}$ order $\Delta \Sigma$ modulator. A power-efficient stochastic signal processor implemented in an area of $\mathbf{0. 1 2} \mathbf{m m}^{2}$ uses $\Delta \Sigma$ modulator output bitstream to learn and compensate artifacts occurring due to concurrent electric stimulation. The proposed system approach is validated on a prototype ASIC in 180 nm technology and consumes a total power of $6.83 \mu \mathrm{~W}$, of which the stochastic signal processor consumes only $0.51 \mu \mathrm{~W}$. Peak-peak stimulation artifacts of $\mathbf{2 0 0 ~ m V}$ are reduced by $\sim 33 \mathrm{~dB}$ over a 10 kHz bandwidth, making it a leading-edge artifact suppression system.