Figure 10

Rotable 3D scatter-plots of first three PC scores. The Figure shows an example automatic clustering of dataset A (fragments A and B are conditions before and after clustering analysis) and automatic classification of dataset B (fragments C and D are conditions before and after clustering). Each dot represents a spike situated in the 3D PC feature space. Four dense clusters are clearly visible and classified. Noise-near waveforms and their PCs located far away from cluster centres, as evaluated by Mahalanobis distance, are marked as red dots.