Neuro-fuzzy recognition of K-complexes in sleep EEG signals

A new pattern recognition procedure is introduced for the automatic detection of K-complexes in EEG sleep polygraphies employing fuzzy logic and neural networks. The developed detection system was implemented in C, supplying a hardware-independent solution with a very rapid recognition of the K-complexes. With the aid of the determined K-complexes and detected sleep spindles, the respective stage 2 in the sleep polygraphy can be evaluated directly from the EEG signal and visualized. This represents a significant criterion for the objective assessment of a patient's sleep quality.