Neural spike sorting with spatio-temporal features

The paper analyses signals that have been measured by brain probes during surgery. First background noise is removed from the signals. The remaining signals are a superposition of spike trains which are subsequently assigned to different families. For this two techniques are used: classic PCA and code vectors. Both techniques confirm that amplitude is the distinguishing feature of spikes. Finally the presence of various types of periodicity in spike trains are examined using correlation and the interval shift histogram. The results allow the development of a visual aid for surgeons