The Study Of ECG-based Respiration Detection Algorithm And Its Application In Sleep Analysis

Sleep apnea syndrome(SAS)is a common sleep disorders,it seriously affects people's sleep quality.In recent years,more and more attention has paid to the researches on the SAS.The traditional method of monitoring sleep apnea syndrome is monitored by polysomnography.Polysomnography can accurately detect the sleep apnea syndrome,but the discomfortableness caused by multiple probes can affect sleep quality which would affect doctors' diagnosis,and the price of monitoring for a long time in care room is too high for people.Therefore,the study of using one channel physiological signal to detect SAS is very important.In this thesis,an electrocardiogram(ECG)-derived respiratory(EDR)algorithm based on ECG R-wave and S-wave,and a sleep apnea detection algorithm base on EDR power spectrum are proposed.Firstly,the algorithm used a median filter,a notch filter and a low-pass filter to remove the baseline drift,power-line interference and electromyographical(EMG)interference,to maintain a high noise-signal ratio.And then,on the basis of threshold method,an adaptive threshold QRS wave detection algorithm is proposed;using the sum of R-wave peak value and S-wave valley value as a new characteristic value to extract the respiratory signal.And the self-adaptive threshold method and Burg power spectrum estimation were used to estimate the respiration rate.Finally,using the sleep apnea events' characteristic in the frequency domain to detect sleep apnea syndrome,analyzing the EDR of different types of sleep apnea syndrome to increase the applicability of this algorithm.By using the MIT-BIH Polysomnographic database and self-collected data(50 volunteers,including 31 men and 19 women,age distribution is 27±8 years,all of them are healthy people),the algorithm is validated.The respiratory rate estimation error of the normal group was lower 2 times per minute than that of sleep apnea syndrome group;in both normal group and SAS group,the EDR signal can follow the change of reference breath,and sleep apnea syndrome could be detected according to the EDR power spectrum.The algorithm is simple and practical,because of the giving up the complex computing methods,and it had a very good application prospects,especially in the field of wearable devices.