Research On The Method Of Heartbeat Separation And Feature Extraction Based On Bio-radar

Heartbeat signal is one of the most important physiological signals which can reveal the abnormal conditions of human health.Common heartbeat signal detection methods is contacted,including Electrocardiograph(ECG)method and Photo Plethysmograph(PPG)method,However,these methods are not suitable for patients with burns,infectious diseases,and electrodes sensitive.Bio-radar can efficiently solve the above problems due to its advantages of non-contact and penetration.The heartbeat signal acquired by bio-radar is weak and easily interfered.Firstly,Electromagnetic wave emitted by bio-radar antenna will be disturbed by environmental noise,resulting in the decrease of Signal-to-Noise Ratio(SNR).Secondly,respiratory harmonics will also affect the detection,resulting in poor separation effect.In addition,the physiological and anatomical characteristics of heart make it worth studying in terms of bio-radar accurate heartbeat signal detection,such as in which detection position and angle can acquire the best signal,and there are few relevant researches at home and abroad currently.In order to solve the above problems,the performance of multiple detection modes of heartbeat signal using bio-radar,the separation and feature extraction of heartbeat signal are studied.The main research work and results are as follows:1.The design and construction of software and hardware platform for heartbeat separation and feature extraction of bio-radar is completed.The principle of bio-radar detecting heartbeat signal is introduced,the hardware platform is constructed,and the upper computer program is developed.2.The heartbeat signal separation method based on bio-radar is studied which is divided into rough separation and sub-separation.Rough separation includes noise removal and preliminary separation of heartbeat and respiration.Firstly,the static clutter is removed by 100-order DC offset removal method.Secondly,the high-frequency interference is removed by a 5Hz low-pass filter.Thirdly,the maximum sum of squares method is used to locate the position of the human.Finally,the heartbeat signal contained respiratory harmonic is obtained by a bandpass filter.The second step is sub-separation.Firstly,the respiratory harmonic frequency is located according to its fundamental frequency.Secondly,the correlation analysis is used to determine which respiratory harmonic to be suppressed.Finally,the feedback notch filter based on parameter optimization is used to suppress respiratory harmonic interference.3.The Physiological characteristics of the heart determine that different detection positions and directions will lead to different signal quality,so the performance of front and back detection mode、frontal multi-angle detection mode and bio-radar,and ECG comparison detection mode are studied before heartbeat signal feature extraction.Front and back detection results show that the heartbeat amplitude of front is superior to that from back.Frontal multi-angle detection results show that the amplitude of heartbeat signal based on radar reaches its maximum at 120° during the process of changing the detection angle from 20° to 160°(with a step of 20°).Comparative studies of radar and ECG at 120° show that radar can detect the motion information of the heart in the blind ECGs segments(Diastasis).By studying the performance of different detection modes of heartbeat,the optimal heartbeat detection position was found.4.Based on the performance of different detection modes,the feature extraction of heart rate variability in time-domain and frequency-domain of heartbeat signal is studied.Firstly,the peak position of heartbeat signal is determined by the comparison method.Secondly,the R-R interval is obtained by the abscissa difference of the peak position.Thirdly,the time-domain characteristics are obtained by calculating the R-R interval.Finally,the frequency domain signal of R-R interval is obtained by Fourier transform after interpolation of R-R interval,and the frequency-domain characteristics are obtained by calculating the frequency domain R-R interval.5.The separation method and feature extraction method are evaluated.The test results show that the separation method in this paper can well suppress the respiratory harmonic,and the average spectrum amplitude suppression rate reaches 85.6%.The feature extraction method well extracts the heart rate variability characteristics of the heartbeat signal,The difference of heart rate variability characteristics between bio-radar and ECG is less than 5%,showing a good consistency.This study has obtained preliminary results in the research of bio-radar heartbeat signal separation and feature extraction.At the same time,further work is needed in the future:1.The sub-separation of heartbeat signal in bio-radar echo,and the notch performance can be improved by adding other sessions to the feedback path of the feedback notch filter;2.In the feature extraction of heartbeat signal in bio-radar echo,further work is needed on the relationship between the heartbeat signal detected by bio-radar and ECG,as well as the correspondence relationship between radar heartbeat features and heart disease.