Automatic Creation of Hypnogram

Abstract

Polysomnography is an important diagnostic method used in sleep monitoring of patients with various sleep disorders. The formulation of a precise diagnosis requires correct evaluation and interpretation of polysomnography EEG and this evaluation is provided by a doctor. In order to save time, various automatic evaluations methods have been developed. One of these methods is based on signal division into 30 second long sections which are subsequently analyzed and reduced using adaptive segmentation and parameters are computed from these segments. Cluster analysis is used to divide individual segments and the class with biggest quantity of segments is called “priority class”. Definition of sleep states is then based on priority class. The end result is a hypnogram and the average success rate of the proposed method is about 68%.