A method for extracting respiratory wave signals based on pulse waves

Abstract

The present invention discloses a method for extracting a respiratory wave signal based on a pulse wave, and belongs to the technical field of respiratory rate parameter monitoring. The present invention proposes a method for merging a respiratory wave with higher reliability based on a respiratory wave extracted from a pulse wave peak and a respiratory wave extracted from a pulse wave trough. The present technology calculates respiratory waves from the pulse wave peak and the pulse wave trough, respectively, and then merges a respiratory wave with higher signal quality through signal quality, peak and trough templates, and correlation coefficients, and then calculates the respiratory rate to improve accuracy.

Description

Technical Field

[0001] The invention belongs to the technical field of respiratory rate parameter monitoring, and in particular relates to a method for extracting respiratory wave signals based on pulse waves. Background Art

[0002] Respiration is a routine monitoring parameter and one of the most important parameters for judging a person's physiological condition. Normal human breathing is not only regular but also uniform. The breathing rate of an adult is 16 to 20 rpm (respiration per minute). The breathing rate of a child is faster than that of an adult, at around 20 to 30 rpm. The breathing rate of a newborn can reach around 44 rpm. In conventional respiratory monitoring, according to the way the respiratory signal is picked up, it can be simply divided into direct measurement method and indirect measurement method.

[0003] 1. Direct measurement method

[0004] According to the difference in the way of obtaining respiratory signals, it can be simply divided into two measurement methods: 1.1

[0006] An airflow detection sensor is placed in the nostril to pick up the airflow (respiratory signal), and the various measurement parameters of breathing are obtained by measuring the exhaled airflow and calculating the periodicity of the airflow change; 1.2

[0008] Impedance or acceleration sensors are placed in the chest or abdomen to pick up fluctuation signals (respiration signals). For example, ECG electrodes with impedance sensors added are placed in the chest or abdomen to collect impedance change signals in the chest or abdomen due to breathing. Alternatively, acceleration sensors are placed in the chest or abdomen to collect acceleration signals caused by the fluctuation of the chest or abdomen during breathing, thereby obtaining respiratory signals. The periodicity of the respiratory signals is then calculated to obtain various measurement parameters of the respiratory signal.

[0009] 2. Indirect measurement method

[0010] A method that does not directly collect the respiratory signal, but derives the respiratory signal through other signals and then calculates the respiratory measurement parameters. 2.1

[0012] The respiratory signal is derived from the ECG signal. Due to the influence of respiratory movement, the ECG axis in the main propagation direction of the heart wave rotates, causing the QRS wave amplitude of the ECG signal to change. The main manifestation is that the QRS wave amplitude decreases during inspiration and increases during exhalation. That is, the respiratory movement has a modulating effect on the QRS wave amplitude of the ECG. Therefore, the respiratory signal can be derived based on the change in the ECG QRS wave amplitude. 2.2

[0014] The respiratory signal is derived from the pulse wave signal. Due to the influence of respiratory movement, the chest or abdomen rises and falls periodically, affecting the negative pressure in the chest and thus the pressure of the ascending aorta. The periodic rise and fall of the chest or abdomen will affect the contraction and relaxation of blood vessels, which is mainly manifested as a decrease in the pulse wave amplitude during inspiration and an increase in the pulse wave amplitude during exhalation. The pulse wave amplitude is modulated, so the respiratory signal can be derived based on the change in the pulse wave amplitude.

[0015] Of the two methods mentioned above, the direct measurement method has a relatively high accuracy, but it also faces many problems. On the one hand, it is necessary to deploy special sensors. Whether it is an airflow detection sensor placed at the nostrils or a sensor placed in the chest or abdomen, it will cause a certain degree of discomfort to the patient. At the same time, the patient's activities will also be restricted, because when the patient is moving, the sensor may fall off or shift, affecting the pickup of the respiratory signal, causing inaccurate measurement results, or even giving completely wrong measurement results. When the patient needs thoracic surgery or chest disinfection, the ECG electrodes or accelerometers need to be removed, and the continuity of respiratory monitoring cannot be guaranteed. In addition, the medical equipment used in the direct measurement method is often large and expensive, and has low applicability in scenarios with high portability requirements.

[0016] In summary, adding a respiratory measurement function without adding sensors and measuring respiratory function through pulse waves is a reliable method with the advantages of high portability, relatively low price, reduced inconvenience to users, and continuous respiratory measurement.

[0017] High-quality respiratory wave extraction through pulse waves is a prerequisite for high-quality respiratory measurement. However, in the existing pulse wave extraction respiratory wave technology, the respiratory wave signal is basically extracted through the amplitude variable of the pulse wave peak or the pulse wave trough. When the pulse wave peak or trough is disturbed, the extracted respiratory wave signal quality is not high and the calculated respiratory parameters are inaccurate. Summary of the invention

[0018] The purpose of the present invention is to provide a method for extracting respiratory wave signals based on pulse waves to overcome the existing problems, so as to extract respiratory waves with high signal quality regardless of whether the pulse wave signal quality is good or the pulse wave peak or pulse wave trough is disturbed.

[0019] The present invention is achieved through the following technical solutions:

[0020] A method for extracting a respiratory wave signal based on a pulse wave comprises the following steps:

[0021] (1) Pulse wave signal filtering to remove the baseline drift and high-frequency interference noise of the pulse wave signal;

[0022] (2) pulse wave peak and trough detection to obtain the pulse wave peak and trough positions;

[0023] (3) calculating the pulse wave peak signal quality index based on the pulse wave peak information;

[0024] (4) calculating the pulse wave trough signal quality index based on the pulse wave trough information;

[0025] (5) Based on the information of pulse wave peak-to-valley distance and valley-to-peak distance, calculate the peak-to-valley distance.

[0026] valley and valley-peak signal quality indices;

[0027] (6) calculating a pulse wave comprehensive signal quality index based on the pulse wave peak signal quality index, the pulse wave trough signal quality index, the peak-to-trough and trough-to-peak signal quality indexes;

[0028] (7) extracting the respiratory wave signal A1 based on the pulse wave peak and the pulse wave peak signal quality index of the above step (3);

[0029] (8) extracting the respiratory wave signal B1 based on the pulse wave trough and the pulse wave trough signal quality index of the above step (4);

[0030] (9) Based on the peak-valley and valley-

[0031] The respiratory wave signal A1 of the above step (7) is corrected to obtain the respiratory wave signal A2. Similarly, the respiratory wave signal B1 of the above step (8) is corrected to obtain the respiratory wave signal B2.

[0032] (10) aligning the respiratory wave signal A2 with the respiratory wave signal B2 based on the pulse wave peak position and the pulse wave trough position;

[0033] (11) detecting the respiratory wave peak and the respiratory wave trough of the respiratory wave signal A2 and the respiratory wave signal B2 respectively, and obtaining the peak and trough positions of the respiratory wave signal A2 and the peak and trough positions of the respiratory wave signal B2 respectively;

[0034] (12) establishing a signal template MA2 for the peaks and valleys of the respiratory wave signal A2, and establishing a signal template MB2 for the peaks and valleys of the respiratory wave signal B2;

[0035] (13) Based on the signal template MA2, a correlation coefficient is calculated for each respiratory wave of the respiratory wave signal A2; based on the signal template MB2, a correlation coefficient is calculated for each respiratory wave of the respiratory wave signal B2;

[0036] (14) Based on the pulse wave comprehensive signal quality index of the above step (6), the pulse wave peak signal quality index, the pulse wave trough signal quality index, the respiratory wave signal A2, the respiratory wave signal B2, and the correlation coefficient calculated in the above step (13), the final respiratory wave signal C is obtained comprehensively.

[0037] Furthermore, the signal quality index in steps (3) to (5) can be obtained by calculating the amplitude changes of the pulse wave peak and the pulse wave trough.

[0038] Furthermore, the extracted respiratory wave signals in step (7) and step (8) are obtained by using a spline fitting method.

[0039] Furthermore, the signal template established in step (12) is obtained by averaging the respiratory wave signal.

[0040] The pulse wave signal is easily disturbed, and when disturbed, it may only disturb the pulse wave peak or the pulse wave trough. Therefore, the respiratory wave signal extracted based on a single pulse wave peak, or the respiratory wave extracted based on a single pulse wave trough is easily disturbed and affected, affecting the accuracy of the respiratory rate. Combining the two respiratory wave signals extracted based on the pulse wave peak and the pulse wave trough, when one respiratory wave is disturbed, the other respiratory wave can be used to improve the quality of the extracted respiratory wave and the accuracy of the respiratory rate. The reason why two respiratory waves can be used to synthesize a more credible respiratory wave is based on three points:

[0041] (1) According to the periodicity and continuity of human breathing, the respiratory wave is periodic and smooth;

[0042] (2) The peak-to-peak spacing of the respiratory wave signal extracted based on the pulse wave peak should be similar to the peak-to-peak spacing of the respiratory wave signal extracted based on the pulse wave trough (since digital sampling cannot achieve analog sampling, it may not be possible to accurately sample the pulse wave peak and pulse wave trough positions, so there will be slight errors and they cannot be exactly the same);

[0043] (3) The peak-to-valley amplitude of the respiratory wave signal extracted based on the pulse wave peak should be similar to the corresponding peak-to-valley amplitude of the respiratory wave signal extracted based on the pulse wave trough;

[0044] Based on the above description, a method for merging a respiratory wave with higher reliability based on a respiratory wave extracted from a pulse wave peak and a respiratory wave extracted from a pulse wave trough is proposed.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] This technology calculates the respiratory wave from the pulse wave peak and pulse wave trough respectively, and then merges the respiratory wave with higher signal quality through signal quality, peak and trough template, and correlation coefficient, and then calculates the respiratory rate to improve accuracy. DETAILED DESCRIPTION

[0047] In order to further explain the present invention, it is described below in conjunction with the following specific embodiments.

[0048] Embodiment 1:

[0049] Extract the respiratory wave through the blood oxygen pulse wave. According to the above, the peak and trough positions of the pulse wave are detected, and then two respiratory waves are extracted based on the peak amplitude and position of the pulse wave and the trough amplitude and position of the pulse wave, and the respiratory wave features are calculated. The merging is started through the respiratory wave features.

[0050] (1) The peak-to-peak distance, peak-to-trough distance, trough distance, and peak-to-trough amplitude of the pulse wave A and the pulse wave B during the nth respiratory wave (n represents any one of the respiratory waves) are similar, and the quality of the two respiratory wave signals is considered to be similar. This is determined based on the peak-to-peak distance or peak-to-trough amplitude of the respiratory wave A and the respiratory wave B and the previous historical respiratory wave distance or respiratory wave amplitude;

[0051] a.

[0052] If the respiratory wave spacing or respiratory wave amplitude is similar, the signal is considered reliable, and the arithmetic mean of respiratory wave A and respiratory wave B is used as the final respiratory wave;

[0053] b.

[0054] The breathing wave spacing and the breathing wave amplitude are not similar, and the correlation coefficient with the peak and valley template of each breathing wave is calculated, and the breathing wave with the largest correlation coefficient is taken as the final breathing wave;

[0055] (2) The nth respiratory wave peak-to-peak distance, respiratory wave peak-to-valley distance, respiratory wave trough distance, and respiratory wave peak-to-valley amplitude of respiratory wave A and respiratory wave B are not similar, and it is considered that the two respiratory waves may be interfered. Calculate the correlation coefficient with each respiratory wave peak-to-valley template, and calculate the correlation coefficient with the most recent final respiratory wave at the same time, and take the respiratory wave with the highest two correlation coefficients as the final respiratory wave.

Claims

1. A method for extracting respiratory wave signals based on pulse waves, characterized in that: The steps include: (1) Pulse wave signal filtering to remove the baseline drift and high-frequency interference noise of the pulse wave signal; (2) Pulse wave peak and trough detection to obtain the pulse wave peak and trough positions; (3) Calculating the pulse wave peak signal quality index based on the pulse wave peak information; (4) calculating the pulse wave trough signal quality index based on the pulse wave trough information; (5) Calculate the peak-to-valley and valley-to-peak signal quality index based on the information of the pulse wave peak-to-valley and valley-to-peak distances; (6) calculating the pulse wave comprehensive signal quality index based on the pulse wave peak signal quality index, the pulse wave trough signal quality index, the peak-to-trough and trough-to-peak signal quality indexes; (7) extracting the respiratory wave signal A1 based on the pulse wave peak and the pulse wave peak signal quality index in step (3); (8) extracting the respiratory wave signal B1 based on the pulse wave trough and the pulse wave trough signal quality index in the above step (4); (9) Based on the peak-to-valley and valley-to-peak signal quality indexes of the above step (5), the respiratory wave signal A1 of the above step (7) is corrected to obtain the respiratory wave signal A2. Similarly, the respiratory wave signal B1 of the above step (8) is corrected to obtain the respiratory wave signal B2; (10) Aligning the respiratory wave signal A2 with the respiratory wave signal B2 based on the pulse wave peak position and the pulse wave trough position; (11) Detecting the respiratory wave peak and respiratory wave trough of the respiratory wave signal A2 and the respiratory wave signal B2 respectively, and obtaining the peak and trough positions of the respiratory wave signal A2 and the peak and trough positions of the respiratory wave signal B2 respectively; (12) A signal template MA2 is established for the peaks and valleys of the respiratory wave signal A2, and a signal template MB2 is established for the peaks and valleys of the respiratory wave signal B2; (13) Based on the signal template MA2, a correlation coefficient is calculated for each respiratory wave of the respiratory wave signal A2; based on the signal template MB2, a correlation coefficient is calculated for each respiratory wave of the respiratory wave signal B2; (14) Based on the pulse wave comprehensive signal quality index of the above step (6), the pulse wave peak signal quality index, the pulse wave trough signal quality index, the respiratory wave signal A2, the respiratory wave signal B2, and the correlation coefficient calculated in the above step (13), the final respiratory wave signal C is obtained by synthesis.

2. The method for extracting respiratory wave signals based on pulse waves according to claim 1, characterized in that: The signal quality index described in step (3) to step (5) is obtained by calculating the amplitude changes of the pulse wave peak and the pulse wave trough.

3. The method for extracting respiratory wave signals based on pulse waves according to claim 1, characterized in that: The extracted respiratory wave signals described in step (7) and step (8) are obtained by using a spline fitting method.

4. The method for extracting respiratory wave signals based on pulse waves according to claim 1, characterized in that: The signal template established in step (12) is obtained by averaging the respiratory wave signal.