A method and medium for adaptive optimization of VME parameters in chest impedance signal processing

Abstract

This invention relates to a method and medium for processing chest impedance signals with adaptive optimization of VME parameters. The method includes the following steps: acquiring a collected chest impedance signal; extracting a baseline signal from the chest impedance signal using a first VME algorithm to obtain a baseline-de-baseline signal, wherein the penalty coefficient of the first VME algorithm is determined through adaptive selection; and performing denoising processing on the baseline-de-baseline signal to extract the respiratory component. Compared with existing technologies, this invention has advantages such as high signal processing efficiency and the ability to process data from different subjects under different conditions.

Description

Technical Field

[0001] This invention relates to the field of chest impedance signal processing technology, and in particular to a chest impedance signal processing method and medium with adaptive optimization of VME parameters. Background Technology

[0002] Impedance pneumography (IP) technology measures chest impedance data and extracts respiratory information for the diagnosis of lung diseases or routine health monitoring. It is a non-invasive and painless measurement method with significant clinical value. To improve the accuracy of diagnostic results, the processing of chest impedance signals and the extraction of respiratory information are becoming increasingly important.

[0003] The chest impedance signal measured by IP devices is often affected by external interference, containing high-frequency noise and baseline drift, components unrelated to respiratory components. Baseline drift is related to the subject's residual lung volume, body posture, and chest muscle tension. Therefore, when the subject is in motion, the baseline of the chest impedance signal is more complex and difficult to effectively fit and remove using polynomials. Thus, using frequency domain methods to remove baseline drift and noise represents a new approach for removing chest impedance baseline drift.

[0004] Empirical Mode Decomposition (EMD) is a time-frequency domain signal processing method that decomposes a signal into different frequency components, called Intrinsic Mode Functions (IMFs), in the frequency domain based on its time-domain characteristics. EMD is suitable for analyzing nonlinear and non-stationary signal sequences, but due to a lack of rigorous mathematical theory, its robustness to noisy signals is weak in practical applications. In 2014, a more stable signal processing algorithm was proposed, called Variational Mode Decomposition (VMD). VMD can also decompose complex signals into multiple components according to their frequency range. Based on VMD, Variational Mode Extraction (VME) was proposed in 2018. Unlike EMD and VMD, VME does not output components across all frequency ranges; instead, it extracts only the component whose center frequency is close to the desired frequency, thus resulting in faster computation.

[0005] VMD and VME have wide applications in biomedicine and other fields, including extracting respiratory components from electrocardiogram signals and gear fault detection. Currently, no one has applied these algorithms to human impedance signal processing. Therefore, it is necessary to research and develop applications of these algorithms in human impedance signal processing to improve the reliability of chest impedance signal acquisition. Summary of the Invention

[0006] The inventors of this invention discovered in their research that VME is very suitable for extracting respiratory signals from thoracic impedance signals, but some key issues need to be addressed:

[0007] First, the VME algorithm requires determining the center frequency ω_d of the extracted signal and the penalty coefficient α that determines the bandwidth of the extracted signal. If the penalty parameter is too large, the extracted signal bandwidth will be too narrow, leading to loss of respiratory signals when used for chest impedance signal processing; if the penalty parameter is too small, other noise components will be mixed into the extracted signal. Currently, there is no method to ensure good extraction results when determining the input parameters; it mostly relies on experience and repeated trials to determine the input parameters.

[0008] Second: Although the frequency bands of the respiratory component, baseline component, and high-frequency noise component in the chest impedance signal are separated in the spectrum for an individual, the frequency ranges of these components are not fixed for different individuals. Therefore, the frequency ranges of different components may overlap between individuals. Thus, when using the VME algorithm to extract the respiratory component, the same set of input parameters will not be effective for all individual measurement signals.

[0009] Third: The bandwidth of the high-frequency component and the baseline component are different from the frequency difference of the respiratory component, so extracting the respiratory component with only one VME is not very effective.

[0010] The purpose of this invention is to overcome the shortcomings of the prior art and provide a VME parameter adaptive optimization method and medium for chest impedance signal processing with better signal processing efficiency.

[0011] The objective of this invention can be achieved through the following technical solutions:

[0012] A method for processing chest impedance signals with adaptive optimization of VME parameters includes the following steps:

[0013] Acquire the collected chest impedance signal;

[0014] The baseline signal is extracted from the chest impedance signal using the first VME algorithm to obtain the baseline-drift-free signal. The penalty coefficient of the first VME algorithm is determined by adaptive selection.

[0015] The baseline-de-signal signal is denoised to extract the respiratory component.

[0016] Furthermore, the center frequency ω of the first VME algorithm d It is 0.

[0017] Furthermore, the penalty coefficient of the first VME algorithm is adaptively selected and determined through the following process:

[0018] Establish an index i, let the penalty coefficient α = 100000 * i, let i range from 1 to 100, use the current VME algorithm to extract the baseline respectively, and calculate the average power P(i) of the signal after removing the baseline;

[0019] Take i as the abscissa and P(i) as the ordinate to draw a graph, connect the point (1, P(1)) and the point (100, P(100)) to form a straight line, and denote the slope of this straight line as k;

[0020] Let i start from 2 and increase by a step of 1, based on the penalty coefficient obtained at each step, use the current VME algorithm to extract the baseline signal, and calculate the average power P(i) of the signal after removing the baseline. Denote the detrended P(i) curve as P′(i);

[0021] When i increases to make P′(i + 1) < P′(i) hold, denote m = i, stop increasing, and record α at this time m = 100000 × m, α m is the optimal α value of the VME algorithm, and then determine the penalty coefficient of the first VME algorithm.

[0022] Furthermore, the expression of the average power P(i) is:

[0023] ​​​​​​​​​​​​​​​​​​​​​​​​​​​

[0031] The present invention also provides an electronic device, comprising:

[0032] One or more processors;

[0033] Memory; and

[0034] One or more programs stored in memory, the programs including instructions for performing the chest impedance signal processing method with adaptive optimization of VME parameters as described above.

[0035] The present invention also provides a computer-readable storage medium including one or more programs executable by one or more processors of an electronic device, said one or more programs including instructions for performing the chest impedance signal processing method of VME parameter adaptive optimization as described above.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. This invention is the first to apply the VME algorithm to human chest impedance data processing to obtain chest impedance change signals caused by respiration. Compared with time-domain algorithms such as polynomial fitting and fixed-threshold frequency-domain algorithms such as bandpass filtering and wavelet transform, the VME algorithm has better performance in processing complex and variable bioimpedance signals. Furthermore, when using the VME algorithm, this invention first uses an adaptively optimized penalty coefficient VME algorithm to extract the baseline signal, and then uses a fixed-parameter VME algorithm to extract the respiratory component. This can accurately and reliably extract the baseline signal from the chest impedance signal, and then extract the respiratory component. It can effectively extract chest impedance changes caused by respiration when processing data from different subjects under different conditions, and also has the advantages of fast calculation speed and high efficiency.

[0038] 2. This invention determines the optimal penalty coefficient by finding the inflection point of a convex curve when the data dispersion is high. Specifically, the inflection point of the convex curve is defined as the point with the longest straight-line distance from the beginning and end points of the curve, and the coordinates of this point are found by using a curve detrending method. This can accurately and reliably achieve adaptive optimization of the penalty coefficient, thereby improving the reliability of impedance signal processing.

[0039] 3. This invention has the advantages of fast calculation speed and high efficiency. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0041] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0042] refer to Figure 1 As shown, this embodiment provides a method for adaptively optimizing chest impedance signals based on VME parameters, including the following steps:

[0043] S1. Acquire the collected chest impedance signal.

[0044] S2, Baseline Removal Step.

[0045] In this embodiment, the first VME algorithm is used to extract the baseline signal from the chest impedance signal to obtain the baseline-drift-free signal. The penalty coefficient of the first VME algorithm is determined by adaptive selection.

[0046] Specifically, the adaptive optimization process of the parameters of the first VME algorithm is as follows:

[0047] a) Determine the input parameters ω of the VME d Since the frequency of the baseline signal is very close to 0, this step directly sets the center frequency ω. d Set it to 0.

[0048] b) Create an index i, and set the penalty coefficient α = 100000*i. Set i to 1 and 100, respectively, extract the baseline using VME, and calculate the average power P(i) of the signal after removing the baseline.

[0049] α = 100000 × i

[0050] {B(x)}=VME({R(x)},α,0)

[0051] {D(x)}={R(x)}-{B(x)}

[0052]

[0053] Where B(x) is the baseline signal, D(x) is the respiratory signal, and R(x) is the raw chest impedance signal. The independent variable x is the discretized measurement time, expressed as:

[0054]

[0055] Where t is the measurement time and fs is the sampling rate.

[0056] c) Plot a graph with i as the x-axis and P(i) as the y-axis, connecting (1, P(1)) and (100, P(100)) with a straight line. The slope k of the straight line is:

[0057]

[0058] d) Let i start from 2 and increment by 1. At each step, use VME to extract the baseline signal and calculate the average power P(i) of the remaining signal. Denote the detrended P(i) curve as P′(i), which is expressed as:

[0059] P′(i) = P(i) - k × i

[0060] e) When i increases to P′(i + 1) < P′(i), record m = i and stop step d). At this time, P′(m) is the highest value point of the P′(i) curve, corresponding to the turning point of the P(i) curve. Record i = m and α m = 100000 × m, which is the optimal α value for the baseline-removed signal. The baseline-removed signal {D m (x)} obtained using this α m value is saved and input to step S3.

[0061] Since the points on the curve in this embodiment are relatively scattered, it is difficult to find the inflection point using the derivative method, and the function itself is an increasing function, so there is no inflection point in the mathematical sense. Therefore, in this embodiment, the inflection point of the convex curve is defined as the point farthest from the straight line connecting the start and end points of the curve, and the method of detrending the curve is used to find the coordinates of this point, determine the optimal penalty parameter, so that VME has better effects when processing complex and variable bio-impedance signals.

[0062] S3. Denoising step.

[0063] In this embodiment, the second VME algorithm is used to denoise the baseline-removed signal and extract the respiration component, which specifically includes the following sub-steps:

[0064] a) Plot the power spectrum of the baseline-removed signal {D m (x)} obtained in step S2. Set the frequency value corresponding to the highest peak point in the power spectrum as the center frequency ω d of the second VME algorithm.

[0065] b) Set the penalty parameter α to 30000 and use VME to extract the respiration signal:

[0066] {S(n)} = VME({D m (x)}, 30000, ω d )

[0067] In other embodiments, the penalty parameter α can also be set to other values greater than 20000.

[0068] If the above methods are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0069] To verify the performance of this invention, experiments were conducted on simulated signals and measured data in this embodiment, and the results were compared with those of bandpass filtering, wpsTID, fixed-parameter VMD, and VME algorithms. wpsTID is another chest impedance signal processing algorithm previously proposed by the inventors' team. Four types of simulated signals were used: short signals with simple frequency and baseline, long signals with complex frequency, long signals with complex baseline, and long signals with complex frequency and baseline. Each signal consists of three parts: baseline, useful signal, and high-frequency noise.

[0070] s(t)=b(t)+f(t)+v

[0071] Where f(t) is the useful signal, b(t) is the baseline signal, and v is high-frequency noise generated using Gaussian noise with a power of 0.01W. The short signal length is 36s, and the long signal length is 480s. The expressions for the baseline and effective signal of the short signal are as follows:

[0072] b1(t)=0.1×t

[0073] f1(t) = 5sin(2π × 0.3 × t)

[0074] The baseline expression for a long frequency composite signal is the same as that for a short signal. The useful signal is divided into three segments, and the expression is as follows:

[0075]

[0076] The expression for the effective signal of a complex long-baseline signal is the same as that for a short-baseline signal f1(t). The baseline expression is:

[0077]

[0078] The baseline expression for long signals with complex frequencies and baselines is the same as that for b3(t), and the expression for effective signals is the same as that for f2(t).

[0079] The processing results are shown in Tables 1 to 4. VME uses fixed input parameters, with a penalty parameter α of 20000 and a center frequency ω. d The Hz frequency is set to 0.2. The number of modes in VMD is set to 3. The passband range of the bandpass filter is set to 0.2 to 0.8 Hz. All other input parameters are set to their default values. This method is referred to as Adaptive VME.

[0080] The measured data consisted of 70 segments of chest impedance signals collected from 35 volunteers, with one segment collected from each volunteer in both resting and active states. In this embodiment, MAE (Mean Absolute Error) and MSE (Mean Square Error) were used as evaluation metrics. Experimental results show that the processing results of this invention are significantly superior to other algorithms.

[0081] Table 1. Processing results of short signals

[0082]

[0083] Table 2. Processing results of frequency composite long signals

[0084]

[0085] Table 3. Processing results of complex long baseline signals

[0086]

[0087] Table 4. Processing results for long signals with complex baselines and frequencies.

[0088]

[0089] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript. These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for processing chest impedance signals with adaptive optimization of VME parameters, characterized in that, Includes the following steps: Acquire the collected chest impedance signal; The baseline signal is extracted from the chest impedance signal using the first VME algorithm to obtain the baseline-drift-free signal. The penalty coefficient of the first VME algorithm is determined by adaptive selection. The baseline-de-signal signal is denoised, and the respiratory component is extracted. The penalty coefficient of the first VME algorithm is adaptively selected and determined through the following process: Create an index i Let the penalty coefficient α = 100000* i ,make i For values ​​of 1 and 100, the baseline is extracted using the current VME algorithm, and the average power of the signal after baseline removal is calculated. ; by i The x-axis is... Plot a graph with the ordinate as the ordinate. Connect the points (1, P(1)) and (100, P(100)) with a straight line. The slope of this line is denoted as . k ; make i Starting from 2 and increasing by 1 step, based on the penalty coefficient obtained at each step, the baseline signal is extracted using the current VME algorithm, and the average power of the signal after removing the baseline is calculated. The trend will be de-trend The curve is denoted as ; when i Growth to At the time of establishment, record m = i Stop incrementing and record the current state. , This is the optimal α value for the VME algorithm, which is then used to determine the penalty coefficient for the first VME algorithm.

2. The method for adaptive optimization of chest impedance signal processing based on VME parameters according to claim 1, characterized in that, The center frequency of the first VME algorithm It is 0.

3. The method for adaptive optimization of chest impedance signals based on VME parameters according to claim 1, characterized in that, The average power The expression is: in, For respiratory signals, { , For the baseline signal, { , The original chest impedance signal, independent variable x For discretized measurement time, This refers to the number of measurement points.

4. The method for adaptive optimization of chest impedance signals based on VME parameters according to claim 1, characterized in that, The and The relationship is represented as: 。 5. The method for adaptive optimization of chest impedance signals based on VME parameters according to claim 1, characterized in that, The baseline-de-denoising signal is denoised using a second VME algorithm. During the denoising process, the center frequency of the second VME algorithm is... Determined in the following ways: Obtain the power spectrum of the baseline-removed signal, and use the frequency value corresponding to the highest peak point in the power spectrum as the center frequency of the second VME algorithm. .

6. The method for adaptive optimization of chest impedance signal processing based on VME parameters according to claim 5, characterized in that, The penalty coefficient of the second VME algorithm is greater than 20000.

7. The method for adaptive optimization of chest impedance signals based on VME parameters according to claim 6, characterized in that, The penalty coefficient for the second VME algorithm is 30000.

8. An electronic device, characterized in that, include: One or more processors; Memory; and One or more programs stored in memory, the one or more programs including instructions for performing the chest impedance signal processing method of adaptive optimization of VME parameters as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, Includes one or more programs executable by one or more processors of an electronic device, said one or more programs including instructions for performing the chest impedance signal processing method of adaptive optimization of VME parameters as described in any one of claims 1-7.

Images