Detection and classification of transient signals. Sorting neural waveforms

The study of biological neural networks requires reliable extracellular recordings that provide data on the collective behavior of neurons. Extracellular recordings contain the transient waveforms generated by several neurons located at the tip of the electrode. These neural spike waveforms have to be detected and classified to determine the firing times of different neurons. Many recording conditions result in high noise levels that increase the difficulty of neural spike sorting. This article describes the generation of the neural waveforms that are recorded, presents a brief survey of methods available for spike sorting, and describes the two methods found to be best. Development of an optimal spike sorting system to provide a valuable tool for neuroscience is in progress at the Applied Physics Laboratory.