Data reduction for real-time enhanced growing neural gas spike sorting with multiple recording channels

Abstract

Neural data analysis algorithms that can rapidly and accurately determine brain states are need to be developed to allow the implementation of closed-loop neural feedback controls for neuroscience studies and neural disorder treatment. In the past, we developed an Enhanced Growing Neural Gas (EGNG) algorithm that can be used to rapidly sort streaming neural spikes in real-time with very limited computational resources, suitable for implementation using digital electronic technologies for system miniaturization. Further development of data reduction is needed to extend the EGNG algorithm to sort neural spikes recorded from a multichannel neural probe. Here, we propose to use two identification methods – peak intensity and area integration – to identify the strongest neural spike among the adjacent channels in order to reduce the size of the recorded data, before sending the neural spikes to the downstream EGNG algorithm for spike sorting. This modification may lead to capability enhancement for the EGNG algorithm to rapidly sort neural spikes recorded from a multi-channel electrode for future closed-loop neural control experiments and treatments.