Independent-component-analysis-based spike sorting algorithm for high-density microelectrode array data processing

Microelectrode arrays (MEAs) become an important tool for neurophysiology research. They are instrumental in revealing neural network formation processes and inter-cell communication schemes, which helps to understand the functioning of the human brain and to treat it´s diseases. The electrode pitch of current CMOS-based MEAs can be as low as 18 ?m, which allows for recording the activity of single cells simultaneously on several channels. Each electrode in turn records the activity of several adjacent neurons. The presented algorithm employs Independent Component Analysis (ICA) method to recover the spike signals and to assign them to a particular neuron. To overcome the fundamental ICA requirement of linearly mixed independent sources, which is not satisfied in the case of neuronal recordings, the algorithm runs in a loop, successively extracts traces with spiking activity, overlays those with previously detected ones and assigns signals to individual neurons.