A novel unsorted spike feature based real-time sensory event detection for closed-loop control

Abstract

The goal of the current study was to investigate the possibility of sensory event detection from the afferent signal recorded by using a multichannel micro-electrode on the dorsal root ganglion without spike sorting. We particularly considered to extract the most informative feature vector from recorded neural signals using feature combination of previous studies such as electroneurograhpy signals, electromyography signals and neural spikes. Principal component analysis was used to reduce dimensionality of feature vector, and multilayer perceptron classifier was used to detect sensory events. In this study, a novel unsorted spike-based feature extraction methods was proposed to enhance the detection accuracy of sensory events.