E. Kaniusas, L. Mehnen, H. Pfützner, B. Saletuand R. Popovic
EVALUATION OF ACOUSTIC SIGNALS FOR THE DETECTION OF SLEEP APNEA EVENTS
The present paper concerns acoustic measurements in connection with the sleep apnea syndrome. The latter induces pathophysiologic changes in the respiratory and circulatory systems and is diagnosed by polysomnography (PSG). PSG comprises the measurement of many parameters and thus tends to be expensive and time consuming. In contrast to available portable devices which use many distributed sensors, we propose a single acoustic sensor in the heart region from which multiple information is derived through excessive signal processing. Measurement is performed by means of a microphone complemented by a stethoscope head. Analysis of the detected signal by means of histogram, cross correlation and FFT shows that it includes information an all three the cardiac activity, the respiratory activity and the snoring sounds. In order to obtain automatic monitoring of obstructive apnea events, a series of signal characteristics in both the time and the spectral domain (e.g., power density, histogram width, PCA components) are subject of processing through adaptive algorithms which are also used for classification. In addition, the method yields breathing patterns and diagnostically relevant physiological data such as changes of the heart rate and the rate of respiration. For the detection of central apnea event, the acoustic sensor is complemented by a magnetostrictive one.