CN108403094A - Method for identifying pulse wave crest - Google Patents

Abstract

A method of identifying a peak of a pulse wave, comprising: (1) collecting waveform signals of pulse waves; (2) checking the main wave frequency distribution; (3) preprocessing an original pulse wave signal to remove noise; (4) selecting a filter based on the frequency range of the main wave, highlighting the main wave by using the selected filter, excluding other peaks, extracting shannon energy envelope from the processed pulse wave signal, and then extracting a local maximum value; (5) and positioning a real pulse wave peak value point on the original pulse wave signal by utilizing the local maximum value. The method disclosed by the invention is accurate in identification when processing signals with high and narrow peaks, has better stability for period division in non-stationary signals, and has strong anti-noise capability for pulse wave signals.

Description

A method for identifying pulse wave peaks

technical field

The present invention relates to a pulse wave waveform identification method, and more particularly, to a method for identifying pulse wave peaks.

Background technique

The human pulse wave generally has 6 characteristic points. As shown in Figure 1, point a represents the opening point of the aortic valve, point c is the highest point of systolic pressure, point d is the point of aortic dilation and pressure reduction, and point e is the point of left ventricle Diastolic starting point, point f is the trough of the diabolic wave, point g is the crest of the diabolic wave, and point b is the end point of the pulse wave. The characteristic points of the pulse diagram respectively reflect the different states of the cardiovascular system and are characteristic points with physiological significance. The pulse map feature points can be used to calculate cardiovascular function parameters, assist in the analysis and judgment of cardiovascular conditions, and can also be used to judge the type and pulse condition of the pulse map. The identification of feature points is of great significance to the automation of pulse diagnosis and is an indispensable part. At present, in the identification of pulse wave feature points, common methods include slope threshold method, wavelet transform, peak height ratio and Gaussian model fitting wrist pulse signal. Due to noise interference and the defects of the method itself, these methods generally have deviations. And the robustness is poor. The main wave is the most obvious and important feature in the pulse wave signal. It is very critical to identify the pulse cycle. Determining the position of the main wave plays an important role in the accurate division of the cycle and the identification of other feature points. Shannon energy envelope has a very good performance in the R-peak detection of ECG signals, and it is also applicable to pulse wave signals. How to use the characteristics of the main wave to accurately identify feature points is an important breakthrough to improve accuracy and robustness.

Contents of the invention

In order to solve the above-mentioned problems in the prior art, the present invention provides a method for identifying pulse wave peaks.

The method for identifying the pulse wave peak provided by the invention comprises:

(1) Gather the waveform signal of the pulse wave;

(2) Check the main wave frequency distribution;

(3) Preprocessing the original pulse wave signal to remove noise;

(4) Select a filter based on the frequency range of the main wave, use the selected filter to highlight the main wave, exclude other peaks, extract the Shannon energy envelope from the processed pulse wave signal, and then extract the local maximum;

(5) Using the local maximum value to locate the real pulse wave peak point on the original pulse wave signal.

Preferably, the method for viewing the frequency distribution of the main wave includes: performing fast Fourier transform on the original pulse wave signal to obtain differentiated signals, and then checking the frequency distribution of the main wave spectrogram.

Preferably, the preprocessing includes using high-pass filtering to remove the baseline drift of the original pulse wave signal, using low-pass filtering to process the glitches in the waveform due to vibration and respiration, and normalizing the pulse wave signal after denoising .

Preferably, the normalized method is: Where d[n] represents the amplitude of the pulse wave at point n, and a[n] is the normalized pulse wave.

Preferably, the calculation method of the Shannon energy envelope is:

Se[n]=-f1 ² [n]log(f1 ² [n])

Among them, F1[n] is the signal after filtering.

Preferably, when extracting the local maximum value, a low-pass filter is used to process the pulse wave signal to reduce the complexity of searching for the local maximum value.

Preferably, the method for extracting the local maximum is selected from one of Hilbert transform, wavelet transform and Fourier transform.

Preferably, the method for extracting the local maximum value is: performing Hilbert transformation on the pulse wave signal, and then using a tangent approximation method to find the peak point of the pulse wave.

Preferably, the method of the Hilbert transform is:

Preferably, the method for locating the real pulse wave peak point on the original pulse wave signal by using the local maximum value includes: on both sides of the local maximum value of the main wave identification curve, looking for A local minimum, where the position of the local minimum corresponds to the start and end of the pulse cycle;

A local maximum value is searched between the start point and the end point, the local maximum value corresponds to the peak of the main wave in the pulse wave waveform signal.

Compared with the prior art, the present invention has the following advantages:

(1) Accurate identification when processing signals with high and narrow peaks;

(2) In the non-stationary signal, the present invention has better stability to the division of period;

(3) Strong anti-noise ability to pulse wave signal.

Description of drawings

Fig. 1 is the waveform signal of human body pulse wave;

Fig. 2 is method flowchart of the present invention;

Fig. 3 is the original pulse waveform signal collected in the embodiment of the present invention;

Fig. 4 is the pulse wave signal after bandpass filtering in the embodiment of the present invention;

Fig. 5 is the Shannon energy envelope extracted in the embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

As shown in Figure 2, the method for identifying pulse wave peak of the present invention comprises:

1. Collect pulse waveform signal;

A pressure-adjustable sensor is placed at the radial artery of the patient's wrist to obtain pulse voltage signals, which are pressurized in sections, converted into digital signals by the pulse acquisition circuit, and stored in the computer, as shown in Figure 3.

2. Check the main wave frequency distribution

Perform fast Fourier transform (FFT) on the original pulse wave signal to obtain the differentiated signal, check the frequency distribution of the main wave spectrogram, and design the following filter based on the frequency range of the main wave;

3. Preprocessing

Use a high-pass filter and a low-pass filter to remove the corresponding noise, including: use a high-pass filter to remove the baseline drift of the original pulse wave signal, and use a 30Hz low-pass filter to deal with the glitches in the waveform due to vibration and breathing. Normalize the pulse wave signal after denoising: where d[n] represents the amplitude of the pulse wave at point n. a[n] is the normalized pulse wave.

4. Find the peak point

According to the frequency range of the main wave, design a suitable band-pass filter (for example, 1-4Hz) to highlight the main wave and exclude other peaks, as shown in Figure 4. Extract the Shannon energy envelope from the processed pulse wave signal, as shown in Figure 5, and then extract the local maximum value, which corresponds to the approximate position of the main wave. Using a low-pass filter can reduce the complexity of searching for local maxima.

The Shannon energy envelope Se[n] is calculated using Se[n]=-f1 ² [n]log(f1 ² [n]),

Among them, F1[n] is the signal after filtering.

There are many ways to find the maximum point of the envelope, not limited to Hilbert transform, wavelet transform and Fourier transform.

Time-domain signals can be extracted from the pulse wave, frequency-domain signals can be extracted through Fourier transform, and time-frequency signal feature points can be extracted through wavelet transform.

An important application of wavelet transform in signal processing is to detect singular points of signals. The rising edge and falling edge of the signal at the singular point correspond to a pair of local extreme values of the wavelet transform detail signal.

The steps of Hilbert transform to determine the local maximum value are as follows:

Use Hilbert transform after low-pass filtering, the method of Hilbert transform is:

The Hilbert transform defines the convolution of 1/πt and x(t).

After the HT, the moving average filter signal H(n) removes low-frequency drift, and locates the peak R(k) of the SEE signal (the pulse wave signal after extracting the Shannon energy envelope) through the zero crossing point from the positive axis to the negative axis. For accurate positioning of the peak point the moving average filter length is very important. The real peak point of the main wave is the maximum point of the pulse signal near R(k) within 0.25s. The onset of each pulse period can be found based on the differentiated signal and the maximum of the dominant wave.

Use the tangent approximation method to find the peak point of the pulse wave, use the minimum value between adjacent peak points on the original pulse wave signal to locate the start point of the pulse wave cycle, and finally place the distance between the two adjacent cycle start points on the original pulse wave signal The maximum value of is determined as the true pulse wave peak point.

Among them, the local detection on the characteristic domain between the adjacent cycle start points on the original pulse wave signal includes:

On both sides of the local maximum value of the main wave identification curve, find the local minimum value on the original pulse wave waveform. The trough v1 of the wave and the trough v2 of the next main wave;

A local maximum value is searched between the starting point and the end point, the local maximum value corresponds to the peak k of the main wave in the pulse waveform signal.

The specific embodiments described above have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention, and are not intended to limit the present invention. Within the spirit and principles of the present invention, any modifications, equivalent replacements, improvements, etc., shall be included in the protection scope of the present invention.

Claims (10)

1. A method for identifying pulse wave peaks, comprising: (1) Gather the waveform signal of the pulse wave; (2) Check the main wave frequency distribution; (3) Preprocessing the original pulse wave signal to remove noise; (4) Select a filter based on the frequency range of the main wave, use the selected filter to highlight the main wave, exclude other peaks, extract the Shannon energy envelope from the processed pulse wave signal, and then extract the local maximum; (5) Using the local maximum value to locate the real pulse wave peak point on the original pulse wave signal. 2. The method according to claim 1, wherein the method for viewing the frequency distribution of the main wave comprises: performing fast Fourier transform on the original pulse wave signal to obtain differentiated signals, and then checking the frequency distribution of the main wave spectrogram. 3. The method according to claim 1, wherein said preprocessing comprises utilizing high-pass filtering to remove baseline drift of the original pulse wave signal, utilizing low-pass filtering to process glitches in the waveform due to vibration and respiration, and removing Normalize the pulse wave signal after noise. 4. The method according to claim 1, wherein the normalized method is: Where d[n] represents the amplitude of the pulse wave at point n, and a[n] is the normalized pulse wave. 5. The method according to claim 1, wherein the calculation method of the Shannon energy envelope is: Se[n]=-f1 ² [n]log(f1 ² [n]) Among them, F1[n] is the signal after filtering. 6. The method according to claim 1, wherein when extracting the local maximum value, a low-pass filter is used to process the pulse wave signal to reduce the complexity of searching for the local maximum value. 7. The method according to claim 1, wherein the method of extracting the local maximum is selected from one of Hilbert transform, wavelet transform and Fourier transform. 8. The method according to claim 1, wherein the method for extracting the local maximum is: performing Hilbert transform on the pulse wave signal, and then using a tangent approximation method to find the peak point of the pulse wave. 9. The method according to claim 8, wherein the method of the Hilbert transform is: 10. The method according to claim 1, wherein the method for locating a real pulse wave peak point on the original pulse wave signal using the local maximum comprises: On both sides of the local maximum value of the main wave identification curve, a local minimum value is found on the original pulse wave waveform, and the position of the local minimum value corresponds to the starting point and the end point of the pulse wave cycle; A local maximum value is searched between the start point and the end point, the local maximum value corresponds to the peak of the main wave in the pulse wave waveform signal.