Detection of the K-Complex in Human EEG Sleep Records.

The paper presents the development of a computer algorithm for the detection of K-complexes in the normal human sleep electroencephalogram EEG. The detection scheme applies time and frequency domain pattern recognition techniques to a single channel central area recording site of digitized EEG data to identify waveforms as K-complexes. The detection of K-complexes resulted in an accuracy of 62.5 7.3 miss error and 30.2 false detections. The primary source of false detections was the identification of Delta wave activity as K-complexes. The detection scheme is also used for scoring sleep EEG data as K-complex Stage 2 or non-K-complex Stages 1 and REM sleep stages. Results of the sleep scoring effort provide an accuracy of 94 when compared to the usually scored EEG. Author