ECG Reconstruction And Heart Rate Variability Analysis Based On Integral Mean Mode Decomposition

Cardiovascular disease(CVD)is a common disease that seriously threatens human health and life.Electrocardiogram(ECG)is an important index commonly used in the examination and diagnosis of CVD.ECG signal is easily polluted by environmental noise.In portable and wearable ECG monitoring systems,environmental noise is more complex and serious,so it is difficult to meet the strict requirements of clinical accurate diagnosis of CVD by using commonly used ECG signal processing algorithms,resulting in a high rate of missed diagnosis and misdiagnosis of CVD.Heart rate variability(HRV)analysis is the only method that can quantitatively analyze autonomic nervous activity and regulatory function in clinic,and it is an important method for clinical diagnosis of CVD.HRV analysis is insensitive to noise and has strong anti-noise ability,so it can effectively evaluate the prevention,diagnosis,treatment and prognosis of CVD.In particular,the quantitative value index of HRV,which can realize the quantitative evaluation of CVD latency,has been paid more and more attention and widely used in clinic.However,the existing HRV linear analysis methods can not fully reveal its rich physiological and pathological information,poor reliability and limited clinical application,while the graphical auxiliary diagnosis method of nonlinear HRV analysis is simple,but the complexity of graphics leads to large errors;the lack of quantitative indicators leads to poor reliability of diagnosis;and the large amount of data required by graphics leads to poor real-time performance.In a word,the above shortcomings not only limit the development of ECG and HRV in the prevention,diagnosis,treatment and prognosis of CVD,but also inhibit the application of portable and wearable ECG equipment.This paper focuses on the nonlinear preprocessing of ECG signals and the quantization of HRV complexity.The main research results of this paper include:1)in order to process ECG and HRV signals effectively,a nonlinear and non-stationary signal processing method based on integral mean mode decomposition(IMMD)is proposed.This method has the characteristic of adaptive orthogonal decomposition,and is suitable for processing not only ECG and HRV data,but also other nonlinear and non-stationary signal data.2)A method of mode component identification and ECG signal reconstruction based on cardiac physical characteristic information is proposed.This method avoids the problem of threshold setting for identifying mode components of QRS characteristic waves,and can suppress the mixed noise of typical wide-band EMG interference and low-frequency baseline drift in ECG.The accuracy and signal-to-noise ratio of ECG signal reconstruction are improved.3)In order to accurately extract HRV data from ECG signals,a R-wave detection and recognition method based on the combination of Hilbert energy envelope and ECG Shannon energy envelope is proposed to detect and locate R peaks accurately.In addition,based on the accurate detection and recognition of R wave,further detection of QRS characteristic waves can also achieve real-time detection of CVD by QRS characteristic waves.4)based on the adaptive mean filtering characteristic of IMMD method,an adaptive multiscale entropy(AMSE)method for quantifying HRV complexity is proposed.Not only the AMSE Samp En value can quantify the HRV complexity,but also the scale itself can indirectly quantify the HRV complexity.This method makes up for the deficiency of fixed scale in multiscale coarse graining method in multi-scale entropy(MSE),and can be used to comprehensively and accurately quantify the complexity of HRV and detect CVD.5)using the Samp En set of mode components as a special multi-scale entropy,the complexity of HRV data is analyzed,which is called intrinsic mode function Samp En(IMFSE).IMFSE can also comprehensively and accurately quantify the complexity of HRV sequences and detect CVD.