Real-time variance based template matching spike sorting system

We present a real-time variance based template matching system for spike sorting. The system employs an ultra-low-power CMOS variance estimator to detect and discriminate between distinct spike classes in real-time. Simulations demonstrate reliable single-unit detection at near 0 dB SNR, and robust classification of action potentials from noisy spike trains; measured results from the variance estimation circuit are in good agreement with simulations. This design thus represents a significant improvement in power efficiency over existing online sorting architectures, which is necessary ill order to achieve fully implantable neural prosthetics.