Heart sound signal denoising method based on improved wavelet threshold

Abstract

The invention belongs to the technical field of computers, particularly relates to a heart sound signal denoising method based on an improved wavelet threshold value, and aims to solve the problem that a heart noise signal component and a heart sound signal component cannot be accurately determined by an IMF component obtained after a heart sound signal is decomposed through CEEMDAN. In the prior art, the selection of a threshold value and a threshold value function in a wavelet threshold value denoising process affects the denoising effect of a signal, when the threshold value is selected too large, some effective components in the heart sound signal are lost, and when the threshold value is selected too small, part of noise in the heart sound signal is not completely eliminated. The method comprises the following steps: decomposing heart sound signals by using CEEMDAN, defining noisy heart sound IMF components and heart sound signal IMF components by using de-trending fluctuation analysis, and de-noising the heart noise IMF components defined by a de-trending fluctuation analysis algorithm by using a wavelet threshold de-noising method of an improved wavelet threshold selection mode. And reconstructing to obtain a de-noised heart sound signal.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a heart sound signal denoising method based on CEEMDAN decomposition and improved wavelet threshold selection, aiming at the problem that the threshold selection is difficult in the wavelet threshold denoising method, the inherent defects of the soft threshold function and the hard threshold function, and the heart sound denoising algorithm of CEEMDAN cannot accurately define the signal-to-noise demarcation point of the IMF component. Background Art

[0002] The heart is an important organ for maintaining human life and is responsible for supplying blood to the body. The heart is composed of atria and ventricles, which push blood to various organs of the body and maintain normal metabolism and function of the human body. The sound produced when the heart contracts and relaxes is the heart sound, which can be divided into the first heart sound (S1), the second heart sound (S2), the third heart sound (S3), and the fourth heart sound (S4). S1 and S2 can be heard under normal circumstances, while S3 and S4 can only be heard faintly in children and adolescents. Among the cases of birth defects in newborns in China, congenital heart disease (CHD for short) ranks first among all congenital diseases. Early detection of newborn CHD and timely treatment can effectively reduce the 20%-30% mortality rate of CHD in infancy. In the "National Comprehensive Prevention and Control Program for Birth Defects", screening for newborn CHD is included in the newborn disease screening program. In the screening of congenital heart disease in newborns, routine heart murmur auscultation is judged by medical staff through a stethoscope. The auscultation results depend on the auscultation skills and experience of medical staff, and cannot avoid the influence of subjective factors. Especially in some remote areas and primary medical units, due to the extreme lack of experienced professional doctors, a considerable number of children with congenital heart disease are missed and delayed in treatment. In recent years, with the emergence of digital stethoscopes, combined with machine learning and pattern recognition technology, the digital collection, storage, analysis of heart sound signals and the intelligent screening of congenital heart disease have become possible. Heart murmur is one of the common manifestations of congenital heart disease. It can indicate abnormalities in heart structure and function. It is an important basis for the diagnosis of congenital heart disease and plays an important role in the classification of heart sounds. The heart sound signals collected by digital stethoscopes are usually interfered by noise, which affects the judgment of heart murmurs. Only by distinguishing the heart murmurs and interfering noises in the heart sound signals of newborns, improving the signal-to-noise ratio of the heart sound signals of newborns by denoising, and retaining the heart murmurs with pathological characteristics in the heart sound signals, can the health status of the newborn heart be accurately assessed.

[0003] In recent years, wavelet threshold denoising algorithm and empirical mode decomposition (EMD) denoising algorithm have been widely used in the noise processing of heart sound signals. Since wavelet transform has good local analysis characteristics for signals in both time domain and frequency domain, selecting an appropriate threshold can better realize the denoising of heart sound signals. However, the non-adaptive selection of traditional thresholds leads to the loss of heart sound information while removing noise. The empirical mode decomposition (EMD) algorithm decomposes the heart sound signal into a set of modal functions arranged from high frequency to low frequency, discards the modal components of the first few orders of high-frequency noise, and reconstructs the remaining modal components to finally obtain the denoised heart sound signal. However, simply discarding components and reconstructing components will lead to the loss of heart sound information and affect the denoising effect. Ensemble empirical mode decomposition (EEMD) algorithm, the EEMD algorithm mixes white noise into the signal to average the intrinsic mode functions of each order, thereby solving the problem of modal aliasing. As for how to define the signal and noise components, EMD and EEMD lack relevant standards, resulting in poor denoising results. The heart sound denoising algorithm based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) cannot accurately define the signal-to-noise demarcation point of the IMF component. Therefore, a heart sound signal denoising method based on improved wavelet threshold is proposed to address the above problems. Summary of the invention

[0004] In order to make up for the deficiencies of the prior art and solve at least one technical problem raised in the background technology, the present invention proposes a heart sound signal denoising method with an improved wavelet threshold.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a heart sound signal denoising method with improved wavelet threshold according to the present invention comprises the following steps:

[0006] S1: Downsample the collected heart sound signal and the heart sound signal obtained online to make the signal frequencies consistent, and perform trend elimination processing to eliminate the trend error caused by the influence of the equipment during collection;

[0007] S2: Perform CEEMDAN decomposition on the noisy heart sound signal and decompose it into different IMF components, making the analysis of the heart sound signal more accurate;

[0008] S3: Detrended fluctuation analysis is used to analyze the trend and fluctuation components of the IMF components decomposed by CEEMDAN, and the scale index α is calculated. The scale index α is used to divide the IMF components into the heart murmur component group (IMFm ) and the heart sound signal component group (IMF u );

[0009] S4: Determine the standard deviation σ of the noise of each component according to the average frequency of the CEEMDAN decomposition component, that is,

[0010]

[0011] Where: f i Smoothly decompose component frequencies of ECG signals;

[0012] The wavelet threshold selection method is improved according to the standard deviation σ, and the threshold function can be selected adaptively according to the number of decomposition layers and the number of components;

[0013] S5: By selecting the appropriate wavelet basis and decomposition layer number for the components containing heart murmurs, a set of wavelet decomposition coefficients is obtained, and improved wavelet threshold denoising is performed to obtain the denoised component group (IMF mm );

[0014] S6: reconstructing the heart murmur component and the heart sound signal component processed by the improved wavelet threshold denoising algorithm;

[0015] Right now

[0016]

[0017] Where: N is the total number of components decomposed by CEEMDAN;

[0018] The final denoised signal is obtained by reconstructing the denoised component group containing the heart murmur signal and the heart sound signal component group.

[0019] Preferably, the improved wavelet threshold selection method in S4 is expressed as:

[0020]

[0021] Where: j is the scale of decomposition, d is the constant coefficient.

[0022] Preferably, the improved wavelet threshold function in S5 is expressed as:

[0023]

[0024] Where: C i,k The kth coefficient of the i-th wavelet decomposition component.

[0025] Beneficial effects of the present invention:

[0026] The present invention provides a heart sound signal denoising method with improved wavelet threshold. Compared with the prior art, the present invention provides a wavelet threshold denoising method based on CEEMDAN decomposition and improved wavelet threshold selection method, which has the following beneficial effects:

[0027] 1. It can not only effectively filter out noise, but also better preserve the detailed characteristics of the heart sound signal and the heart murmur with pathological characteristics. While improving the signal-to-noise ratio of the signal, it maintains the integrity and accuracy of the heart sound waveform.

[0028] 2. The denoising effect of the wavelet threshold denoising algorithm depends on the selection of the wavelet threshold and the threshold function. The empirical mode decomposition has mode aliasing, and the decomposition of the IMF component of the adaptive noise complete set empirical mode decomposition is difficult to determine. The detrended fluctuation analysis algorithm is used to determine the dividing point between the heart murmur component and the heart sound signal component in the IMF component, and the heart sound signal denoising algorithm with an improved wavelet threshold is used to denoise each group of components, thereby improving the computational efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings described herein are used to provide further understanding of the present invention and constitute a part of this application. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0030] In the attached picture:

[0031] Figure 1 is a flow chart of the method of the present invention;

[0032] Figure 2 It is a comparison chart of specific applications of the invented method. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] Specific examples are given below.

[0035] Reference Figure 1 , a heart sound signal denoising method based on improved wavelet threshold mainly includes the following steps:

[0036] Step 1: Downsample the collected heart sound signal and the heart sound signal obtained from the MIT-BIH database to make the signal frequencies consistent, and perform trend elimination processing to eliminate the trend error caused by the influence of the equipment during acquisition.

[0037] Step 2: Perform CEEMDAN decomposition on the noisy heart sound signal to decompose the heart sound signal into different IMF components, making the analysis of the heart sound signal more accurate.

[0038] Step 3: Detrended fluctuation analysis is used to analyze the trend and fluctuation components of the IMF components decomposed by CEEMDAN, and the scale index is calculated. The scale index is used to divide the IMF components into a heart murmur component group and a heart sound signal component group.

[0039] Step 4: Determine the standard deviation of the noise of each component based on the average frequency of the CEEMDAN decomposition components, and then improve the wavelet threshold selection method based on the standard deviation.

[0040] Step 5: Perform wavelet transform on the components containing heart murmur by selecting appropriate wavelet basis and decomposition layer number to obtain a set of wavelet decomposition coefficients, and perform improved wavelet threshold denoising to obtain a denoised component group.

[0041] Step six, reconstruct the heart murmur component and the heart sound signal component processed by the improved wavelet threshold denoising algorithm to obtain the final denoised signal.

[0042] In view of the deficiencies of the prior art, the present invention provides a heart sound signal denoising method based on CEEMDAN decomposition and improved wavelet threshold selection, which solves the problems raised in the above-mentioned background technology. The method first uses CEEMDAN to decompose the noisy heart sound signal into different IMF components, and uses detrended fluctuation analysis to divide the IMF components into heart murmur components and heart sound signal components. Then, for the heart murmur component group, denoising is performed using a wavelet threshold denoising algorithm with improved wavelet threshold selection. Finally, the processed heart murmur signal component group and the heart sound signal component group are reconstructed to obtain the final denoised heart sound signal. The wavelet threshold denoising method with improved wavelet threshold selection in this paper effectively solves the influence of inaccurate wavelet threshold selection on the denoising effect, and the detrended fluctuation analysis effectively solves the problem of inaccurate selection of the demarcation point for distinguishing the IMF components obtained after CEEMDAN decomposition.

[0043] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A heart sound signal denoising method using an improved wavelet threshold, characterized in that: The following steps are involved: S1: Downsample the collected heart sound signal and the heart sound signal obtained online to make the signal frequencies consistent, and perform trend elimination processing to eliminate the trend error caused by the influence of the equipment during collection; S2: Perform CEEMDAN decomposition on the noisy heart sound signal and decompose it into different IMF components, making the analysis of the heart sound signal more accurate; S3: Detrended fluctuation analysis is used to analyze the trend and fluctuation components of the IMF components decomposed by CEEMDAN, and the scale index α is calculated. The scale index α is used to divide the IMF components into the heart murmur component group (IMF m ) and the heart sound signal component group (IMF u ); S4: Determine the standard deviation σ of the noise of each component according to the average frequency of the CEEMDAN decomposition component, that is, Where: f i Smoothly decompose component frequencies of ECG signals; The wavelet threshold selection method is improved according to the standard deviation σ, and the threshold function can be selected adaptively according to the number of decomposition layers and the number of components; S5: By selecting the appropriate wavelet basis and decomposition layer number for the components containing heart murmurs, a set of wavelet decomposition coefficients is obtained, and improved wavelet threshold denoising is performed to obtain the denoised component group (IMF mm ); S6: reconstructing the heart murmur component and the heart sound signal component processed by the improved wavelet threshold denoising algorithm; Right now Where: N is the total number of components decomposed by CEEMDAN; The final denoised signal is obtained by reconstructing the denoised component group containing the heart murmur signal and the heart sound signal component group.

2. The heart sound signal denoising method based on improved wavelet threshold according to claim 1, characterized in that: The improved wavelet threshold selection method in S4 is expressed as: Where: j is the scale of decomposition, d is the constant coefficient.

3. The heart sound signal denoising method of improved wavelet threshold according to claim 1, characterized in that: The improved wavelet threshold function in S5 is expressed as: Where: C i,k The kth coefficient of the i-th wavelet decomposition component.

Images