Detección automática de arousals en registros polisomnográficos

Abstract

The study and analysis of polysomnographic recordings is one of the main tools used for the diagnosis and treatment of sleep illnesses and disorders. One step of that analysis is the manual scoring by a specialist of different events that happen while the patient sleeps. The arousals, which are defined as abrupt changes in the EEG frequency, are one of those events. In this paper we describe a procedure for the automatic detection and scoring of arousals, which consists of three stages: feature extraction, detection and post-processing. In particular, we compare three different detectors: Bayesian, multilayer perceptron (MLP) and support vector machines (SVM). The performance of these detectors is evaluated using polysomnographic recordings taken from several patients.