Robust, on-line, real time system and method for neuron spike recovery

Stanford researchers at the Moore Lab have developed an algorithm for on-line, real time post processing of large amounts of neuronal data from high-density, multi-channel electrophysiological recordings to identify which neurons were firing (on-line spike recovery). The researchers were able to demonstrate on-line spike recovery recorded from hundreds of neurons on hundreds of channels simultaneously using convolutional compressed sensing for sparse signal recovery. The proposed algorithm demonstrated accurate and robust signal recovery comparable to manual spike sorting despite continuously updated measurements. Additionally, the proposed algorithm inherently allows for resolution of spatio-temporal overlapping activity. In combination with an offline spike-sorting initialization, this algorithm provides the means necessary for utilization of single-cell activities in online applications such as closed-loop neuroscience or brain-machine interfaces.

Stage of Research