A method and device for fusion extraction of pulse wave and electrocardiogram signal high-quality waveform

Abstract

The present invention provides a method and device for fusion extraction of pulse wave and electrocardiogram signal high-quality waveforms, the method comprising: 1. acquisition of photoelectric volumetric pulse wave signal PPG and electrocardiogram signal ECG; 2. signal noise reduction and baseline drift elimination preprocessing; 3. PPG and ECG signal fusion feature point detection; 4. multi-feature value extraction; 5. signal waveform morphology and threshold judgment. The present invention automatically extracts high-quality waveforms of PPG and ECG signals in the same time period by real-time acquisition of PPG and ECG signals of the human buttocks and fusion processing, eliminates abnormal segments in the signal, and improves the utilization rate of the signal. The extracted high-quality signal waveform is suitable for identity recognition, and can also be used for the detection of physiological parameters such as blood pressure, heart rate, and blood oxygen.

Description

Technical Field

[0001] The present invention relates to a method and device for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals, belonging to the field of photoelectric signal processing. The extracted high-quality signal waveforms are suitable for identity recognition and can also be used for non-invasive detection of physiological parameters such as daily blood pressure and heart rate. Background Art

[0002] High-quality signal waveforms can accurately extract characteristic parameters, and are the basis for establishing mathematical and physical models and combining machine learning algorithms for identity recognition and physiological parameter detection. They can distinguish users and measure human health status.

[0003] Common physiological signals include photoplethysmography (PPG) and electrocardiogram (ECG). Photoelectric sensors are used to collect data in real time. The amount of light absorbed by the blood in the arteries changes periodically with the blood volume in the blood vessels, and can also reflect the cardiovascular state. The ECG signal reflects the periodic changes in the bioelectricity of various parts of the body during each cardiac cycle. PPG and ECG can be used for research on identity recognition technology, and can also be used for non-invasive monitoring of physiological parameters such as blood oxygen saturation, heart rate, respiratory rate, and blood pressure. However, due to the influence of factors such as ambient light, human motion, and sensor position, the collected physiological signals will produce problems such as noise and baseline drift.

[0004] It is necessary to perform preprocessing such as filtering and noise reduction on the original physiological signals to quickly and accurately extract the characteristic parameters of the signal waveform.

[0005] In recent years, physiological signal noise reduction methods include single use of wavelet threshold noise reduction, empirical mode decomposition (EMD), etc., which are prone to incomplete noise filtering, modal aliasing and endpoint effects, and cannot completely solve the baseline drift problem. Traditional signal processing methods detect feature points separately for PPG signals and ECG signals, with fixed window length and step length. The waveform features of the two signals in the same period of time are not combined for analysis and judgment, and the calculation speed is slow, which cannot effectively eliminate abnormal segments in the signal.

[0006] Based on the above problems, it is of great significance to propose a fusion extraction method for high-quality waveforms of pulse wave and ECG signal. Summary of the invention

[0007] The purpose of the present invention is to provide a method and device for fusion extraction of pulse wave and electrocardiogram signal high-quality waveform, so as to realize the extraction of pulse wave and electrocardiogram signal high-quality waveform in the same time period.

[0008] In order to achieve the above purpose and overcome the existing technical defects, the technical solution adopted is:

[0009] In a first aspect, a method for fusing and extracting high-quality waveforms of a pulse wave and an electrocardiogram signal is provided, comprising:

[0010] Obtain a section of photoelectric volumetric pulse wave (PPG) signal and electrocardiogram (ECG) signal;

[0011] The acquired photoelectric volumetric pulse wave (PPG) signal and electrocardiogram (ECG) signal are preprocessed (noise reduction and baseline drift elimination) to obtain preprocessed PPG signal and ECG signal;

[0012] The pre-processed PPG signal and ECG signal are fused with feature point detection to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal. The window threshold size and step size are changed according to the peak position of the PPG signal to exclude abnormal segments in the signal, and the peak and trough feature points of the PPG and ECG signals and the continuous m cycles of PPG and ECG signal waveform data containing the feature points are obtained.

[0013] Extract characteristic values ​​of single-cycle PPG and ECG signal waveform data to obtain characteristic values ​​of single-cycle PPG and ECG signals;

[0014] Based on the peak and trough feature points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals in each cycle, the morphology and threshold value of the PPG and ECG signal waveforms of m consecutive cycles are judged to obtain high-quality signal waveforms that meet the conditions.

[0015] In some embodiments, the preprocessing includes noise reduction and elimination of baseline drift;

[0016] Further, in some embodiments, preprocessing the acquired photoplethysmography (PPG) signal includes:

[0017] The PPG signal is subjected to bandpass filtering combined with the decomposition layer K PPG =4 variational mode decomposition (VMD) is used for processing.

[0018] Furthermore, in some embodiments, preprocessing the acquired ECG signal includes:

[0019] The ECG signal is decomposed by the number of layers K ECG =5 VMD decomposition and modal components (IMF S ) is used for wavelet transform noise reduction.

[0020] In some embodiments, the fusion feature point detection of the pre-processed PPG signal and ECG signal includes:

[0021] (1) Detect the first peak position of the PPG signal According to the experimental determination of the PPG adjacent wave peak position interval threshold range D, (sampling frequency fs) distance range extracts the maximum value, that is, the second PPG peak P 2 , (m+1)D is used as the window threshold for extracting the peaks of PPG and ECG for m consecutive cycles, and is the step length, save P 1 , P 2 The horizontal and vertical coordinates are: is the horizontal coordinate of the ith peak of the PPG signal, is the ordinate of the ith peak of the PPG signal, sampling frequency fs;

[0022] (2) is the step length, Extract the maximum value in the distance range, i.e. the peak R of the ECG 1 , and determine whether the peak value meets the threshold condition. If it meets the condition, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0023] (3) is the step length, The distance range extracts the minimum value, i.e. the trough S of the ECG 1 , and determine whether the horizontal coordinate position of the trough is greater than the ECG peak R 1 Position, if it meets the requirement, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0024] (4) is the step length, The minimum value of the distance range is extracted, that is, the trough V of PPG 1 , and determine whether the extracted PPG trough value meets the threshold condition. If it meets the threshold condition, save V 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0025] (5) Extract the second ECG peak R 2 , ECG trough S 2 and PPG Trough V 2 The time step is updated to Find the maximum range and update it to By analogy, we get P i (i=1,2,…,m+2), R i (i=1,2,…,m+1),S i (i=1,2,…,m+1), V i The horizontal and vertical coordinate values ​​of (i=1, 2, ..., m+1), as well as the continuous m cycles of PPG and ECG waveform data containing the detection points.

[0026] In some embodiments, feature value extraction is performed on single cycle PPG and ECG signal waveform data, including:

[0027] Extract feature values ​​from single-cycle PPG and ECG signal waveform data, including:

[0028] By detecting the peaks and troughs of the PPG signal for m consecutive cycles, the PPG peak P in each cycle is extracted according to the set interval. i The four points on the left of P i1 , P i2 , P i3 , P i4 , the three points on the right side of the PPG peak i5 , P i6 , P i7 ,PPG Trough V i The four points on the left of V i1 、V i2 、V i3 、V i4 ;

[0029] By detecting the peaks and troughs of the ECG signal for m consecutive cycles, the ECG peak R is extracted according to the set interval. i The two points on the left of R i1 , R i2 , ECG trough S i The three points on the left of S i1 , S i2 , S i3 .

[0030] In some embodiments, the interval is set to

[0031] In some embodiments, based on the peak and trough characteristic points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals of each cycle, the waveforms of the PPG and ECG signals of m consecutive cycles are subjected to morphology and threshold judgment, including:

[0032] Based on the peak and trough characteristic points of PPG and ECG signals and the characteristic values ​​of PPG and ECG signals in each cycle, three slopes are calculated for a single cycle PPG signal, namely point P i With P i4 , P i With P i7 、V i With V i4 The slope between Three slopes are calculated for a single cycle ECG signal, namely point R i With R i2 , R i With Si2 , S i With S i2 The slope between The variance of the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals of m consecutive cycles is calculated, and the calculated variance is compared with the set variance threshold. If the variance threshold condition is not met, the feature point detection step is returned to repeat the fusion extraction process of the pulse wave and ECG signal high-quality waveform until the variance threshold conditions are met and the PPG and ECG signal high-quality waveforms are obtained.

[0033] In some embodiments, the variance is calculated for the peak and trough characteristic points, characteristic values ​​and slopes of the PPG and ECG signals of m consecutive cycles, and the calculation formula is as follows:

[0034]

[0035] Where m is the number of signal cycles, X i Represents the peak and trough amplitude, characteristic value point amplitude and slope corresponding to the i-th period signal, It represents the average value of the peaks, troughs, amplitudes and slopes of the characteristic value points of the signal for m consecutive periods.

[0036] In some embodiments, m=5.

[0037] In a second aspect, a device for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals is provided, comprising:

[0038] A signal acquisition module is configured to acquire a photoelectric volumetric pulse wave (PPG) signal and an electrocardiogram (ECG) signal;

[0039] A preprocessing module is configured to preprocess the acquired photoelectric volume pulse wave (PPG) signal and electrocardiogram (ECG) signal respectively to obtain a preprocessed PPG signal and ECG signal;

[0040] A feature point detection module is configured to detect the fused feature points of the preprocessed PPG signal and ECG signal to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal, change the window threshold size and step size according to the peak position of the PPG signal, exclude abnormal segments in the signal, and obtain the peak and trough feature points of the PPG and ECG signals and the waveform data of the PPG and ECG signals of continuous m cycles containing the feature points;

[0041] The feature value extraction module is configured to extract feature values ​​from the single-cycle PPG and ECG signal waveform data to obtain the single-cycle PPG and ECG signal feature values;

[0042] The calculation and judgment module is configured to perform morphology and threshold judgment on the PPG and ECG signal waveforms of m consecutive cycles based on the peak and trough characteristic points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals in each cycle, so as to obtain a high-quality signal waveform that meets the conditions.

[0043] In a third aspect, a device for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals is provided, including a processor and a storage medium;

[0044] The storage medium is used to store instructions;

[0045] The processor is used to operate according to the instructions to execute the steps of the method described in the first aspect.

[0046] According to a third aspect, a storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method according to the first aspect are implemented.

[0047] The beneficial effects of the above technical solution adopted in the present invention are as follows: the present invention provides a method for fusion extraction of high-quality waveforms of pulse wave and electrocardiogram signal, by analyzing the pulse wave signal and electrocardiogram signal collected in real time, targeted noise reduction and elimination of baseline drift processing are performed, PPG and ECG signals are fused to detect the peaks and troughs of the signal waveform, and characteristic values ​​are extracted; and morphology and threshold judgment are performed to automatically exclude abnormal segments in the signal, extract high-quality signal waveforms, and improve the accuracy of feature parameter extraction and signal utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a general flow chart of a method for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals.

[0049] Figure 2 It is a schematic diagram of pulse wave and ECG signal feature point detection.

[0050] Figure 3 It is a feature point detection flow chart of the fusion extraction method of pulse wave and electrocardiogram signal high-quality waveform. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0052] Example 1

[0053] A method for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals, comprising:

[0054] Obtain a section of photoelectric volumetric pulse wave (PPG) signal and electrocardiogram (ECG) signal;

[0055] The acquired photoelectric volumetric pulse wave (PPG) signal and electrocardiogram (ECG) signal are preprocessed (noise reduction and baseline drift elimination) to obtain preprocessed PPG signal and ECG signal;

[0056] The pre-processed PPG signal and ECG signal are fused with feature point detection to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal. The window threshold size and step size are changed according to the peak position of the PPG signal to exclude abnormal segments in the signal, and the peak and trough feature points of the PPG and ECG signals and the continuous m cycles of PPG and ECG signal waveform data containing the feature points are obtained.

[0057] Extract characteristic values ​​of single-cycle PPG and ECG signal waveform data to obtain characteristic values ​​of single-cycle PPG and ECG signals;

[0058] Based on the peak and trough feature points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals in each cycle, the morphology and threshold value of the PPG and ECG signal waveforms of m consecutive cycles are judged to obtain high-quality signal waveforms that meet the conditions.

[0059] In some embodiments, the preprocessing includes noise reduction and baseline drift elimination.

[0060] Further, in some embodiments, preprocessing the acquired photoplethysmography (PPG) signal includes:

[0061] The PPG signal is subjected to bandpass filtering combined with the decomposition layer K PPG =4 variational mode decomposition (VMD) is used for processing.

[0062] Furthermore, in some embodiments, preprocessing the acquired ECG signal includes:

[0063] The ECG signal is decomposed by the number of layers K ECG =5 VMD decomposition and modal components (IMF S ) is used for wavelet transform noise reduction.

[0064] In some embodiments, the fusion feature point detection of the pre-processed PPG signal and ECG signal includes:

[0065] (1) Detect the first peak position of the PPG signal According to the experimental determination of the PPG adjacent wave peak position interval threshold range D, (sampling frequency fs) distance range extracts the maximum value, that is, the second PPG peak P 2 , (m+1)D is used as the window threshold for extracting the peaks of PPG and ECG for m consecutive cycles, and is the step length, save P1 , P 2 The horizontal and vertical coordinates are: is the horizontal coordinate of the ith peak of the PPG signal, is the ordinate of the ith peak of the PPG signal, sampling frequency fs;

[0066] (2) is the step length, Extract the maximum value within the distance range, i.e. the peak R of the ECG 1 , and determine whether the peak value meets the threshold condition. If it meets the condition, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0067] (3) is the step length, The distance range extracts the minimum value, i.e. the trough S of the ECG 1 , and determine whether the horizontal coordinate position of the trough is greater than the ECG peak R 1 Position, if it meets the requirement, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0068] (4) is the step length, The minimum value of the distance range is extracted, that is, the trough V of PPG 1 , and determine whether the extracted PPG trough value meets the threshold condition. If it meets the threshold condition, save V 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window;

[0069] (5) Extract the second ECG peak R 2 , ECG trough S 2 and PPG Trough V 2 The time step is updated to Find the maximum range and update it to By analogy, we get P i (i=1,2,…,m+2), R i (i=1,2,…,m+1),S i (i=1,2,…,m+1), V i The horizontal and vertical coordinate values ​​of (i=1, 2, ..., m+1), as well as the continuous m cycles of PPG and ECG waveform data containing the detection points.

[0070] In some embodiments, feature value extraction is performed on single cycle PPG and ECG signal waveform data, including:

[0071] Extract feature values ​​from single-cycle PPG and ECG signal waveform data, including:

[0072] By detecting the peaks and troughs of the PPG signal for m consecutive cycles, the PPG peak P in each cycle is extracted according to the set interval. i The four points on the left of P i1 , P i2 , P i3 , P i4 , the three points on the right side of the PPG peak i5 , P i6 , P i7 ,PPG Trough V i The four points on the left of V i1 、V i2 、V i3 、V i4 ;

[0073] By detecting the peaks and troughs of the ECG signal for m consecutive cycles, the ECG peak R is extracted according to the set interval. i The two points on the left of R i1 , R i2 , ECG trough S i The three points on the left of S i1 , S i2 , S i3 .

[0074] In some embodiments, the interval is set to

[0075] In some specific embodiments, the PPG peak P in each cycle is extracted by detecting the peaks and troughs of the PPG signal of m consecutive cycles. i The four points on the left of P i1 =P i -10, P i2 =P i -20, P i3 =P i -30, P i4 = =P i -40, the three points on the right of the PPG peak i5 =P i +10, P i6 =P i +20, P i7 =P i +30, PPG trough V i The four points on the left of V i1 =V i -10, V i2 =V i -20, V i3 =V i -30, V i4= = V i -40;

[0076] By detecting the peaks and troughs of the ECG signal for m consecutive cycles, the ECG peak R is extracted. i The two points on the left of R i1 =R i -10, R i2 =R i -20, ECG trough S i The three points on the left of S i1 =S i -10, S i2 =S i -20, S i3 =S i -30.

[0077] In some embodiments, based on the peak and trough characteristic points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals of each cycle, the waveforms of the PPG and ECG signals of m consecutive cycles are subjected to morphology and threshold judgment, including:

[0078] Based on the peak and trough characteristic points of PPG and ECG signals and the characteristic values ​​of PPG and ECG signals in each cycle, three slopes are calculated for a single cycle PPG signal, namely point P i With P i4 , P i With P i7 、V i With V i4 The slope between Three slopes are calculated for a single cycle ECG signal, namely point R i With R i2 , R i With S i2 , S i With S i2 The slope between The variance of the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals of m consecutive cycles is calculated, and the calculated variance is compared with the set variance threshold. If the variance threshold condition is not met, the feature point detection step is returned to repeat the fusion extraction process of the pulse wave and ECG signal high-quality waveform until the variance threshold conditions are met and the PPG and ECG signal high-quality waveforms are obtained.

[0079] In some embodiments, the variance is calculated for the peak and trough characteristic points, characteristic values ​​and slopes of the PPG and ECG signals of m consecutive cycles, and the calculation formula is as follows:

[0080]

[0081] Where m is the number of signal cycles, Xi Represents the peak and trough amplitude, characteristic value point amplitude and slope corresponding to the i-th period signal, It represents the average value of the peaks, troughs, amplitudes and slopes of the characteristic value points of the signal for m consecutive periods.

[0082] In some embodiments, m=5.

[0083] like Figure 1 As shown, the present invention proposes a method for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals, comprising the following steps:

[0084] Step 1: Collect a pulse wave signal and an ECG signal in real time;

[0085] Step 2: Preprocess the signal in step 1 by noise reduction and baseline drift elimination;

[0086] Step 3: Detect feature points of the signals in step 2, i.e., the pre-processed PPG and ECG signals;

[0087] Step 4: Extract feature values ​​from the signal in step 3;

[0088] Step 5: Perform morphology and threshold judgment on the extracted PPG and ECG signal waveforms of m (m=5) consecutive cycles to obtain high-quality signal waveforms that meet the conditions.

[0089] The pulse wave signal and the electrocardiogram signal are collected by the photoelectric sensor and the electrode sheet on the smart toilet seat. In the process of collecting signals, they are affected by the collection environment, the movement of the tester, electromagnetic interference, etc., and problems such as noise and baseline drift occur, which produce abnormal fluctuation segments and affect the detection of feature points and the extraction of feature values. In view of these problems, the present invention first performs noise reduction and baseline drift elimination preprocessing on the pulse wave signal and the electrocardiogram signal.

[0090] The present invention firstly performs bandpass filtering on the PPG signal by combining the decomposition layer number K PPG =4 variational mode decomposition (VMD) is used to process the ECG signal through K ECG =5 VMD decomposition and modal components (IMF S ) is used for wavelet transform noise reduction.

[0091] For the PPG signal, the high-frequency noise interference is first filtered out by a Butterworth bandpass filter, and the main pulsation components within 0.4-7Hz are retained. The PPG signal is further decomposed by variational mode decomposition (VMD) to obtain the IMF of each modal component. sThe baseline drift and residual noise in the signal are eliminated by processing. The decomposition level K of VMD directly affects the processing effect of the signal. In the process of signal decomposition, as the decomposition level K increases, the center frequency of each component gradually tends to be stable, which is one of the criteria for determining the K value. In addition, the energy leakage value is used to select the appropriate K value.

[0092] The PPG signal containing N sampling points is recorded as F(t)={F 1 ,F 2 ,F 3 ,…,F N}, the decomposed components are recorded as f j (t) = {f j1 ,f j2 ,f j3 ,…,f jN}, j = 1, 2, ..., K, then the energy of the signal F(t) and the component is:

[0093]

[0094]

[0095] Due to the IMF S The components are approximately orthogonal. If a suitable K value is selected, the energy leakage after signal decomposition is small:

[0096]

[0097] The number of PPG decomposition layers is determined as K by the center frequency and energy leakage value. PPG =4, the actual noise content of the signal is different or the K value is different. The correlation coefficient between each modal component after decomposition and the original signal is calculated, and the components with correlation coefficient values ​​greater than 0.3 are reconstructed to obtain the PPG signal after down-regulation preprocessing.

[0098] For ECG signals, VMD is first used to decompose the signal. Through simulation analysis, the decomposition layer number K is determined by combining the modal component center frequency and energy leakage value obtained by ECG signal decomposition. ECG =5, which can be determined according to the actual noise content of the signal. The ECG signal is decomposed by VMD to obtain IMF 1 -IMF 5 The baseline drift component in the signal is removed, and the remaining components are denoised and reconstructed again by wavelet transform to obtain the preprocessed ECG signal.

[0099] After the PPG and ECG signals are preprocessed by noise reduction and baseline drift elimination, the pulse wave and ECG signal fusion are further used to detect feature points. The specific steps are as follows:

[0100] Pulse wave and ECG signal fusion feature point detection Figure 2 As shown in the figure, the feature point detection flow chart of the fusion extraction method of pulse wave and ECG signal high-quality waveform is as follows Figure 3 As shown:

[0101] (1) Detect the first peak position of the PPG signal According to the experimental determination of the PPG adjacent wave peak position interval threshold range D, (Sampling frequency) distance range extracts the maximum value, that is, the second PPG peak P 2 , (m+1)D is used as the window threshold for extracting the peaks of PPG and ECG for m consecutive cycles, and is the step length, save P 1 , P 2 The horizontal and vertical coordinate values.

[0102] (2) is the step length, Extract the maximum value in the distance range, i.e. the peak R of the ECG 1 , and determine whether the peak value meets the threshold condition. If it meets the condition, save R 1 If the horizontal and vertical coordinates do not match, return to step (1) and move the PPG window.

[0103] (3) is the step length, The distance range extracts the minimum value, i.e. the trough S of the ECG 1 , and determine whether the horizontal coordinate position of the trough is greater than the ECG peak R 1 Position, if it meets the requirement, save R 1 If the horizontal and vertical coordinates do not match, return to step (1) and move the PPG window.

[0104] (4) is the step length, The minimum value of the distance range is extracted, that is, the trough V of PPG 1 , and determine whether the extracted PPG trough value meets the threshold condition. If it meets the threshold condition, save V 1 If the horizontal and vertical coordinates do not match, return to step (1) and move the PPG window.

[0105] (5) Extract the second ECG peak R 2 , ECG trough S 2 and PPG Trough V 2 The time step is updated to Find the maximum range and update it to By analogy, we get P i (i=1,2,…,m+2), R i (i=1,2,…,m+1),S i(i=1,2,…,m+1), V i The horizontal and vertical coordinate values ​​of (i=1, 2, ..., m+1), as well as the continuous m cycles of PPG and ECG waveform data containing the detection points.

[0106] Based on the peaks and troughs of the PPG and ECG signals of m consecutive cycles obtained by fusion detection of pulse wave and ECG signal, feature values ​​are further extracted. i is the reference point, and the interval is At the peak P i Take four points P on the right i1 , P i2 , P i3 , P i4 , take three points P on the left i5 , P i6 , P i7 ; Take the trough V i is the reference point, and the interval is Take four points V on the right side of the trough in turn i1 、V i2 、V i3 、V i4 For a single cycle of PPG signal, the peak R i is the reference point, and the interval is At the peak R i Take two points R on the right i1 , R i2 , with the trough S i is the reference point, and the interval is In the trough S i Take three points S on the right i1 , S i2 , S i3 .

[0107] In some embodiments, the PPG peak P in each cycle i The four points on the left of P i1 =P i -10, P i2 =P i -20, P i3 =P i -30, P i4 = =P i -40, the three points on the right of the PPG peak i5 =P i +10, P i6 =P i +20, P i7 =P i +30, PPG trough V i The four points on the left of Vi1 =V i -10, V i2 =V i -20, V i3 =V i -30, V i4 = = V i -40; ECG peak R i The two points on the left of R i1 =R i -10, R i2 =R i -20, ECG trough S i The three points on the left of S i1 =S i -10, S i2 =S i -20, S i3 =S i -30.

[0108] Finally, the pulse wave and ECG signal are fused and detected with feature points and extracted feature values: i , P i1 , P i2 , P i3 , P i4 , P i5 , P i6 , P i7 、V i 、V i1 、V i2 、V i3 、V i4 , R i , R i1 , R i2 , S i , S i1 , S i2 , S i3 . Calculate the three slopes of the PPG signal, The three slopes of the ECG signal, For PPG and ECG signals of m consecutive cycles, a single cycle of PPG and ECG signals includes 20 feature points and 6 slopes for variance threshold judgment:

[0109] The variance is calculated for the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals for m consecutive cycles. The calculation formula is as follows:

[0110]

[0111] Where m is the number of signal cycles, X i Represents the peak and trough amplitude, characteristic value point amplitude and slope corresponding to the i-th period signal, It represents the average value of the peaks and valleys, characteristic value points and slopes of the signal for m consecutive cycles;

[0112] Finally, the variance threshold is judged for the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals of m consecutive cycles (m=5). If the variance threshold condition is not met, the feature point detection step is returned to repeat the fusion extraction process of the pulse wave and ECG signal high-quality waveforms. In response to the variance threshold conditions being met, high-quality waveforms of the PPG and ECG signals are obtained.

[0113] Finally, the PPG and ECG feature point values ​​and high-quality signal waveform data that meet the requirements are output.

[0114] In summary, a method for fusion extraction of high-quality waveforms of pulse wave and electrocardiogram signal of the present invention performs noise reduction and baseline drift elimination processing on the collected original electrocardiogram signal and pulse wave signal respectively, detects feature points by fusing the preprocessed signal with the pulse wave and the electrocardiogram signal, changes the window threshold and step size in real time for feature point detection of PPG and ECG signals, accurately extracts peak and trough points of m consecutive cycles in the same time period, further extracts feature values ​​based on the peak and trough feature points, and finally performs variance threshold judgment on the extracted 20 feature points and 6 slopes, effectively eliminates abnormal fluctuation segments in the signal waveform, and improves signal utilization.

[0115] Example 2

[0116] In a second aspect, a device for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals is provided, comprising:

[0117] A signal acquisition module is configured to acquire a photoelectric volumetric pulse wave (PPG) signal and an electrocardiogram (ECG) signal;

[0118] A preprocessing module is configured to preprocess the acquired photoelectric volume pulse wave (PPG) signal and electrocardiogram (ECG) signal respectively to obtain a preprocessed PPG signal and ECG signal;

[0119] A feature point detection module is configured to detect the fused feature points of the preprocessed PPG signal and ECG signal to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal, change the window threshold size and step size according to the peak position of the PPG signal, exclude abnormal segments in the signal, and obtain the peak and trough feature points of the PPG and ECG signals and the waveform data of the PPG and ECG signals of continuous m cycles containing the feature points;

[0120] The feature value extraction module is configured to extract feature values ​​from the single-cycle PPG and ECG signal waveform data to obtain the single-cycle PPG and ECG signal feature values;

[0121] The calculation and judgment module is configured to perform morphology and threshold judgment on the PPG and ECG signal waveforms of m consecutive cycles based on the peak and trough characteristic points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals in each cycle, so as to obtain a high-quality signal waveform that meets the conditions.

[0122] Example 3

[0123] In a second aspect, this embodiment provides a pulse wave and electrocardiogram signal high-quality waveform fusion extraction device, including a processor and a storage medium;

[0124] The storage medium is used to store instructions;

[0125] The processor is used to operate according to the instructions to execute the steps of the method according to embodiment 1.

[0126] Example 4

[0127] In a third aspect, this embodiment provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in Example 1 are implemented.

[0128] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0129] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0130] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0131] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0132] The above is only used to illustrate the main implementation scheme of the present invention. The present invention is described in detail based on specific examples. Those skilled in the art can further improve and change the content of the technical solution based on the guidance of the specific principles and implementation steps of the above content. The protection scope of the present invention is applicable to the claims and their equivalent modifications.

Claims

1. A method for extracting high-quality waveforms of pulse wave and electrocardiogram signal by fusion. It is characterized in that include: Obtain a section of photoelectric volumetric pulse wave (PPG) signal and electrocardiogram (ECG) signal; Preprocessing the acquired photoplethysmogram (PPG) signal and electrocardiogram (ECG) signal respectively to obtain preprocessed PPG signal and ECG signal; The pre-processed PPG signal and ECG signal are fused with feature point detection to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal. The window threshold size and step size are changed according to the peak position of the PPG signal to exclude abnormal segments in the signal, and the peak and trough feature points of the PPG and ECG signals and the continuous m cycles of PPG and ECG signal waveform data containing the feature points are obtained. The method of fusion feature point detection of the pre-processed PPG signal and ECG signal includes: (1) Detect the first peak position of the PPG signal According to the experimental determination of the PPG adjacent wave peak position interval threshold range D, The maximum value is extracted from the distance range, i.e. the second PPG peak P 2 , (m+1)D is used as the window threshold for extracting the peaks of PPG and ECG for m consecutive cycles, and is the step length, save P 1 , P 2 The horizontal and vertical coordinates are: is the horizontal coordinate of the ith peak of the PPG signal, is the ordinate of the ith peak of the PPG signal, sampling frequency fs; (2) is the step length, Extract the maximum value in the distance range, i.e. the peak R of the ECG 1 , and determine whether the peak value meets the threshold condition. If it meets the condition, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (3) is the step length, The distance range extracts the minimum value, i.e. the trough S of the ECG 1 , and determine whether the horizontal coordinate position of the trough is greater than the ECG peak R 1 Position, if it meets the requirement, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (4) is the step length, The minimum value of the distance range is extracted, that is, the trough V of PPG 1 , and determine whether the extracted PPG trough value meets the threshold condition. If it meets the threshold condition, save V 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (5) Extract the second ECG peak R 2 , ECG trough S 2 and PPG Trough V 2 The time step is updated to Find the maximum range and update it to By analogy, we get P i (i=1,2,…,m+2), R i (i=1,2,…,m+1),S i (i=1,2,…,m+1), V i The horizontal and vertical coordinate values ​​of (i=1, 2, ..., m+1), and the continuous m cycles of PPG and ECG waveform data containing the detection point; The characteristic values ​​of the single-cycle PPG and ECG signal waveform data are extracted to obtain the characteristic values ​​of the single-cycle PPG and ECG signals, including: By detecting the peaks and troughs of the PPG signal for m consecutive cycles, the PPG peak P in each cycle is extracted according to the set interval. i The four points on the left of P i1 , P i2 , P i3 , P i4 , the three points on the right side of the PPG peak i5 , P i6 , P i7 ,PPG Trough V i The four points on the left of V i1 、V i2 、V i3 、V i4 ; By detecting the peaks and troughs of the ECG signal for m consecutive cycles, the ECG peak R is extracted according to the set interval. i The two points on the left of R i1 , R i2 , ECG trough S i The three points on the left of S i1 , S i2 , S i3 ; Based on the peak and trough feature points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals in each cycle, the waveforms of the PPG and ECG signals of m consecutive cycles are judged by morphology and threshold to obtain high-quality signal waveforms that meet the conditions, including: Based on the peak and trough characteristic points of PPG and ECG signals and the characteristic values ​​of PPG and ECG signals in each cycle, three slopes are calculated for a single cycle PPG signal, namely point P i With P i4 , P i With P i7 、V i With V i4 The slope between Three slopes are calculated for a single cycle ECG signal, namely point R i With R i2 , R i With S i2 , S i With S i2 The slope between The variance of the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals of m consecutive cycles is calculated, and the calculated variance is compared with the set variance threshold. If the variance threshold condition is not met, the feature point detection step is returned to repeat the fusion extraction process of the pulse wave and ECG signal high-quality waveform until the variance threshold conditions are met and the PPG and ECG signal high-quality waveforms are obtained.

2. A method for extracting pulse wave and electrocardiogram signal high-quality waveform fusion as claimed in claim 1, It is characterized in that The preprocessing of the acquired pulse wave PPG signal includes: The PPG signal is subjected to bandpass filtering combined with the decomposition layer K PPG =4 is used for the variational mode decomposition.

3. A method for fusion extraction of pulse wave and electrocardiogram signal high-quality waveform as claimed in claim 1, It is characterized in that The preprocessing of the acquired ECG signal includes: The ECG signal is decomposed by the number of layers K ECG =5 and the wavelet transform denoising of each modal component is used for processing.

4. A method for fusion extraction of pulse wave and electrocardiogram signal high-quality waveform as claimed in claim 1, It is characterized in that The interval is set to 5. A method for fusing and extracting high-quality waveforms of pulse waves and electrocardiogram signals as claimed in claim 1, It is characterized in that m＝5。 6. A fusion extraction device for pulse wave and electrocardiogram signal high-quality waveform, It is characterized in that include: A signal acquisition module is configured to acquire a photoelectric volumetric pulse wave (PPG) signal and an electrocardiogram (ECG) signal; A preprocessing module is configured to preprocess the acquired photoelectric volume pulse wave (PPG) signal and electrocardiogram (ECG) signal respectively to obtain a preprocessed PPG signal and ECG signal; A feature point detection module is configured to detect the fused feature points of the preprocessed PPG signal and ECG signal to obtain the peaks and troughs of the PPG signal and the peaks and troughs of the ECG signal, change the window threshold size and step size according to the peak position of the PPG signal, exclude abnormal segments in the signal, and obtain the peak and trough feature points of the PPG and ECG signals and the waveform data of the PPG and ECG signals of continuous m cycles containing the feature points; The method of fusion feature point detection of the pre-processed PPG signal and ECG signal includes: (1) Detect the first peak position of the PPG signal According to the experimental determination of the PPG adjacent wave peak position interval threshold range D, The maximum value is extracted from the distance range, i.e. the second PPG peak P 2 , (m+1)D is used as the window threshold for extracting the peaks of PPG and ECG for m consecutive cycles, and is the step length, save P 1 , P 2 The horizontal and vertical coordinates are: is the horizontal coordinate of the ith peak of the PPG signal, is the ordinate of the ith peak of the PPG signal, sampling frequency fs; (2) is the step length, Extract the maximum value in the distance range, i.e. the peak R of the ECG 1 , and determine whether the peak value meets the threshold condition. If it meets the condition, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (3) is the step length, The distance range extracts the minimum value, i.e. the trough S of the ECG 1 , and determine whether the horizontal coordinate position of the trough is greater than the ECG peak R 1 Position, if it meets the requirement, save R 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (4) is the step length, The minimum value of the distance range is extracted, that is, the trough V of PPG 1 , and determine whether the extracted PPG trough value meets the threshold condition. If it meets the threshold condition, save V 1 The horizontal and vertical coordinates are taken; if they do not match, return to step (1) and move the PPG window; (5) Extract the second ECG peak R 2 , ECG trough S 2 and PPG Trough V 2 The time step is updated to Find the maximum range and update it to By analogy, we get P i (i=1,2,…,m+2), R i (i=1,2,…,m+1),S i (i=1,2,…,m+1), V i The horizontal and vertical coordinate values ​​of (i=1, 2, ..., m+1), and the continuous m cycles of PPG and ECG waveform data containing the detection point; The feature value extraction module is configured to extract feature values ​​from the single-cycle PPG and ECG signal waveform data to obtain the single-cycle PPG and ECG signal feature values; specifically, it includes: extracting the PPG peak P in each cycle according to the set interval by detecting the peaks and troughs of the PPG signal of m consecutive cycles. i The four points on the left of P i1 , P i2 , P i3 , P i4 , the three points on the right side of the PPG peak i5 , P i6 , P i7 ,PPG Trough V i The four points on the left of V i1 、V i2 、V i3 、V i4 ; By detecting the peaks and troughs of the ECG signal for m consecutive cycles, extract the ECG peak R according to the set interval i The two points on the left of R i1 , R i2 , ECG trough S i The three points on the left of S i1 , S i2 , S i3 ; The calculation and judgment module is configured to perform morphology and threshold judgment on the PPG and ECG signal waveforms of m consecutive cycles based on the peak and trough feature points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals of each cycle, and obtain a high-quality signal waveform that meets the conditions, specifically including: based on the peak and trough feature points of the PPG and ECG signals and the characteristic values ​​of the PPG and ECG signals of each cycle, a single cycle PPG signal calculates three slopes, which are point P i With P i4 , P i With P i7 、V i With V i4 The slope between Three slopes are calculated for a single cycle ECG signal, namely point R i With R i2 , R i With S i2 , S i With S i2 The slope between The variance of the peak and trough feature points, feature values ​​and slopes of the PPG and ECG signals of m consecutive cycles is calculated, and the calculated variance is compared with the set variance threshold. If the variance threshold condition is not met, the feature point detection step is returned to repeat the fusion extraction process of the pulse wave and ECG signal high-quality waveform until the variance threshold conditions are met and the PPG and ECG signal high-quality waveforms are obtained.

7. A fusion extraction device for pulse wave and electrocardiogram signal high-quality waveform, It is characterized in that including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 5.

8. A storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images