Research On Fusion Of Multimodal Physiological Signals For Heart Rate Based On Quality Assessment

The usage of accurate heart rate(HR)monitoring in intensive care unit(ICU)and wearable ECG monitoring is very important.Acquiring HR is usually achieved by analyzing electrocardiogram(ECG).However,the contamination of ECG signal obtained by ICUs or wearable devices is very common,and accurate HR acquisition becomes more difficult,which may lead to erroneous estimation of HR,which may endanger the patient’s life safety or cause alarm fatigue of the guardian.This study fuses ECG signal and ABP signal based on their signal quality index(SQI)to improve the accuracy of HR monitoring in ICU,the quality assessment and HR estimation model of dynamic ECG is also established.In this paper,we have made in-depth research on three questions:single-channel HR estimation method,fusion of ECG and ABP to obtain HR and its accuracy evaluation method,dynamic ECG signal quality evaluation and HR estimation method.The main research contents are as follows:(1)Research on single channel HR estimation method.For the ECG signal and the ABP signal,a two-step process of quality assessment and heart rate estimation was required.The gqrs and sqrs algorithms were used to detect R wave peak of the ECG signal.The wabp algorithm was used to detect the starting point of the ABP pulse,and the initial HR was calculated according to the positions of the extracted feature points.The initial HR was processed by Kalman filter to obtain the HR for fusion.The three SQIs of the ECG,i.e.bSQI,sSQI,and kSQI were calculated according to waveform characteristics such as the positions of the R wave peaks of the ECG.Use the fuzzy logic algorithm according to the waveform characteristics of ABP such as the starting point of the pulse to calculate the two SQIs of the ABP signal:jSQI and wSQI.The final SQIs of the ECG and ABP used for fusion can be separately calculated according to the above SQIs.(2)Research on the fusion of ECG and ABP to obtain HR and its accuracy assessment methods.HR estimated from ECG and ABP.respectively,were fused using a SQI-based approach.In order to evaluate the accuracy of HR estimation after fusion,HR estimation results of the single-channel aind fusion were compared with the HR estimation results after expert labeling,and different evaluation indexes,i.e.the HR estimation error and the mean and the standard deviation(SD)of the absolute value of the error were used to verify the effectiveness of the single-channel HR detector and fusion method.(3)A quality assessment model of the dynamic ECG signal is established,and the result of the algorithm evaluation is used to verify the result of the manual evaluation to determine the quality of the ECG signal collected by the smart wearable device.HR estimated from the dynamic ECG signal should be compared with the expert annotated HR to obtain the HR estimation error of the algorithm.The research results show that the proposed fusion of ECG and ABP for HR estimation based on SQI can improve the accuracy of HR estimation in ICU and an effective dynamic ECG signal quality assessment and HR estimation model is also established.