An independent-component-analysis-based time-space processor for the identification of neural stimulation sources

A new scheme for enhancing both compression and interpretability of multichannel biopotentials acquired from implanted neural sensors is presented. It uses spatial and temporal correlation between samples to find patterns associated to external stimulation producing an effect on the sensed tissues, which can be interpreted as sensations or intentions of the subject. Temporal analysis includes spike detection and sorting. Spatial processing is based on an independent component analysis (ICA) of the observed neural activity. Demonstration through modeling and simulation is made of the capability of the system to associate the observed potentials to external stimulations.