Hi-index | 0.00 |
We describe a modification of a growing grid neural net for the purpose of sorting neuronal spike waveforms from extracellular recordings in the central nervous system. We make use of the fact, that real neurons exhibit a refractory period after firing an action potential during which they can not create a new one. This information is utilized to control the growth process of a growing grid, which we use to classify spike waveforms. The new algorithm is an alternative to a standard self-organizing map used in our previously published spike sorting system. Using simulated data, we show that this modification can further improve the accuracy in sorting neuronal spike waveforms.