Discrete-time automatic spike detection circuit for neural recording implants

A circuit for automatic detection of action potentials (spikes) in neural recording devices is presented. This circuit, designed to operate in discrete time, provides an analogue threshold level based on standard deviation of the analogue samples of the input neural signal. Functionality of the circuit was tested for normal prerecorded neural signals as the input. To examine whether the spike detection threshold level tracks low-frequency baseline variations (e.g. local field potentials), the input neural signal was superposed on both a step function and a low-frequency sine wave. Designed in a 0.18 um CMOS technology, according to simulations the circuit consumes 47 W from a 0.9 V supply. A prototype version of the circuit was implemented and successfully tested.