Non-contact blood pressure measuring method combining visible light vision and millimeter wave radar

Abstract

The invention belongs to the technical field of blood pressure detection, and particularly provides a non-contact blood pressure measuring method combining visible light vision and millimeter wave radar. Transmitting a transmitting signal to the vein of the wrist of the human body by using a radar, and collecting echo data; meanwhile, a multispectral camera is used for collecting video signals of the exposed part of the human body. A radar signal and a human body video signal are preprocessed, pulse wave signals are extracted from the two kinds of signals respectively, and the blood pressure is estimated through the subtle difference between pulse waves of different wave bands and the correlation between the multiple pulse waves. According to the detection method, non-oppression and non-contact non-inductive continuous measurement of the blood pressure of the central artery is achieved through the millimeter wave radar and visible light vision, the skin of a living body is not oppressed, blood pressure changes caused by oppression are avoided, and the complexity of blood pressure measurement is obviously reduced.

Description

Technical Field

[0001] The invention relates to the field of biomedicine, and in particular to a non-contact continuous blood pressure measurement system using vision and millimeter wave radar. Background Art

[0002] Blood pressure is a key indicator for assessing the basic vital signs of an organism. It not only reflects the health of the cardiovascular system, but also plays an important role in medical diagnosis, health assessment, and prevention of cardiovascular diseases. Timely and accurate monitoring of blood pressure is essential for improving the early detection rate of cardiovascular diseases and promoting medical care. Studies have shown that continuous monitoring of blood pressure levels and their changes can more sensitively and reliably predict the risks associated with blood pressure than traditional fixed-point measurements. Therefore, dynamic continuous blood pressure monitoring and home self-measurement of blood pressure are more recommended as routine blood pressure monitoring methods. Although the invasive arterial catheter method is regarded as the gold standard for blood pressure measurement, it is complicated to operate, prone to infection, and not suitable for daily monitoring. Currently, commonly used non-invasive blood pressure monitoring methods include auscultation and oscillometric methods, but these methods require the cuff to be inflated and deflated, which may interfere with blood flow and lead to inaccurate measurement results, and also bring inconvenience to patients. Therefore, there is an urgent need for a technical means of non-invasive and continuous monitoring of central arterial blood pressure to reduce patient discomfort and obtain accurate blood pressure information with high temporal resolution.

[0003] In recent years, with the rapid development of sensor and computer technology, it has become possible to measure blood pressure using non-contact methods. One widely studied method is to use high-resolution radar to measure weak arterial disturbances and calculate the pulse wave transit time (PTT) in order to estimate blood pressure using the blood pressure dynamic equation (MK-EE equation). PTT refers to the blood flow time from the aorta to other arteries when the heart beats, which is closely related to blood pressure. However, it is usually difficult to accurately calculate the PTT parameter from a single-channel ECG signal, and multi-channel signal acquisition increases the difficulty of actual operation. Another non-contact method is photoplethysmography (rPPG), which uses camera imaging to collect signals, collects pulse wave signals through optical principles, and models them according to the physiological information contained in the pulse wave (such as blood oxygen, heart rate, blood pressure, etc.), thereby obtaining a continuous blood pressure signal. The rPPG signal is usually a multispectral signal covering different wavelengths of light. However, the feasibility of single-channel blood pressure estimation based on this method is low. Although the fusion of multi-channel signals can improve accuracy, the fusion process is complicated and may lead to signal loss and accuracy loss. Although some methods have achieved non-invasive continuous blood pressure monitoring, these methods all require contact or even compression of the biological skin, causing discomfort to the subjects, and are not suitable for patients with skin diseases, skin burns, etc. In addition, many methods are only applicable to blood pressure monitoring of peripheral arteries. However, since peripheral arteries contain more smooth muscles, peripheral blood vessels are prone to contraction and relaxation changes that are unrelated to blood pressure. It has been proven that the blood pressure of arteries close to the heart (such as the carotid artery) is more closely related to cardiovascular function. Therefore, how to achieve non-contact and non-invasive continuous measurement of central arterial blood pressure through non-compression and non-contact technical means has become an urgent problem to be solved. Summary of the invention

[0004] The technical problem to be solved by the present invention is to realize a non-contact, convenient, continuous and accurate blood pressure measurement system by combining visible light vision and millimeter wave radar to measure blood pressure.

[0005] The technical solution adopted by the present invention is as follows: a non-contact blood pressure measurement method combining visible light vision and millimeter wave radar, the method comprising:

[0006] Step 1: Use a multispectral camera to collect human video signals and select the best region of interest (ROI) for tracking, use a millimeter-wave radar to send a transmission signal to the pulse of the biological body, collect echo data, and use a continuous electronic blood pressure meter to measure real-time continuous blood pressure; the multispectral camera resolution is at least 720p, the frame rate range is set to 15-30fps according to the pulse wave frequency, the collection site is not limited, and the best region of interest is the skin area in the video;

[0007] Step 2: Convert the video signal into a time series signal of each channel;

[0008] Step 3: Extract the millimeter wave radar IQ two-way signal;

[0009] A signal with a large rate of change is selected for subsequent processing, and this time series signal is a type 2 signal;

[0010] Step 4: Preprocess the class 1 signal and class 2 signal of each channel to remove non-pulse wave components. The specific operations include denoising, drift removal, DC component removal, and abnormal point removal.

[0011] Step 5: Extract signal correlation features;

[0012] The three channels of the class 1 signal and the one channel of the class 2 signal are combined into 4-channel data, and the half-width of the pulse wave and the rising area of ​​the heart systolic period of the 4-channel data are extracted. Then, the features extracted from each channel are subtracted and compared with each other to obtain the first type of signal correlation features; the systolic peaks, troughs and the maximum slope points of the pulse waves of each channel are located, and the time differences and amplitude differences of the above feature points between each channel are directly calculated to obtain the second type of signal correlation features;

[0013] Step 6: Construct a blood pressure estimation model and establish a blood pressure measurement system;

[0014] Repeat the above 6 steps to extract different sample features and obtain samples. A blood pressure estimation model is constructed based on all sample features and real-time blood pressure measurement values. The input of the blood pressure estimation model is the two types of signal correlation features extracted in step 5, and the output is the blood pressure value. Use the estimation model to establish and train a blood pressure measurement system.

[0015] Step 7: Use the trained blood pressure measurement system to measure the blood pressure of the new target.

[0016] Furthermore, the method of step 2 is:

[0017] Step 2.1: Divide the acquired image into three channels, namely, images of red spectral band, green spectral band and blue spectral band;

[0018] Step 2.2: After channel separation, the average pixel grayscale value in the best region of interest of each channel image is calculated respectively, and the calculated value is used as the amplitude value of the corresponding channel signal at that moment. This time series signal is a class 1 signal.

[0019] Furthermore, the blood pressure estimation model constructed in step 6 is a multiple linear regression model, and a training set is formed using a part of the features in step 5 and the real-time blood pressure values ​​to obtain the regression parameters of the regression model.

[0020] Furthermore, the calculation method of the average pixel grayscale value in step 2.2 is:

[0021] The average pixel gray value calculation formula is:

[0022] Where n is the number of image frames, N is the total number of pixels, X(n) represents the average pixel grayscale value of the nth frame image, Sn represents the ROI corresponding to the nth frame image, and f(r,c) represents the pixel value at the pixel point (r,c).

[0023] The present invention innovatively combines a multispectral camera and a millimeter-wave radar to collect pulse wave signals. The multispectral camera is used to obtain pulse wave signals in multiple optical bands, while the millimeter-wave radar provides another unique signal collection method, further enriching the data dimension. By utilizing the subtle differences between pulse waves in different bands and the correlation between multiple pulse waves, the system can estimate blood pressure more accurately. This method not only enhances the independence of the information carried in the pulse wave signals corresponding to different wavelengths of light and reduces information loss, but also by combining data from different wavelengths of light and millimeter-wave radar, it can more effectively suppress motion interference and environmental noise that occur in the measurement, thereby significantly improving the accuracy and reliability of blood pressure measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the present invention.

[0025] Figure 2 Device for acquiring video signals, radar signals and reference blood pressure.

[0026] Figure 3 It is a feature that can be extracted from the pulse wave signal of a single channel.

[0027] Figure 4 It is the time difference characteristic and amplitude difference characteristic between the pulse wave signals of the two channels. DETAILED DESCRIPTION

[0028] The present invention adopts an innovative blood pressure measurement method, that is, collecting pulse wave signals to measure blood pressure by combining a multi-spectral camera with a millimeter-wave radar. Compared with the traditional method of simultaneously collecting photoplethysmography (PPG) signals and electrocardiogram (ECG) signals, or collecting PPG signals from multiple parts at the same time, the method of the present invention only needs to combine the data of the multi-spectral camera and the millimeter-wave radar, which not only simplifies the complexity of the blood pressure measurement operation and realizes non-invasive, non-contact, and high-precision blood pressure measurement, but also improves the accuracy and efficiency of blood pressure measurement by integrating the advantages of different technologies.

[0029] The specific implementation is as follows:

[0030] Step 1: Collect human video signals and radar echo data, and select the best ROI for tracking. In a bright environment, use a multispectral camera to collect human video signals, and use a millimeter-wave radar to send a transmission signal to the pulse of the organism and collect echo data. The subjects work and study normally in front of the camera to ensure that the video signal and radar data are not blocked. Select parameters such as video resolution, frame rate, and video acquisition time according to actual needs. The optimal ROI should contain large skin pixels, and the proportion of skin pixels should be no less than 95%. At the same time, use an electronic blood pressure meter to measure real-time continuous blood pressure as a blood pressure reference value.

[0031] Step 2: Generate multi-spectral original pulse wave signal and radar original pulse wave signal. Separate multiple spectral channels (such as RGB channels) of the video image, calculate the average pixel grayscale value in the ROI area of ​​each channel, and obtain the amplitude value of the time series signal to generate a multi-spectral original pulse wave signal. At the same time, extract the signal with obvious changes from the radar echo data as the radar original pulse wave signal.

[0032] The average pixel gray value calculation formula is:

[0033] Where n is the number of image frames, N is the total number of pixels, X(n) represents the average pixel grayscale value of the nth frame image, Sn represents the ROI corresponding to the nth frame image, and f(r,c) represents the pixel value at the pixel point (r,c).

[0034] Step 3: Preprocess the original pulse wave signal, including multispectral and radar data. First, the same preprocessing process is used for the pulse wave signals of the three RGB channels of the multispectral and the original pulse wave signals of the radar. (Only the specific parameters are different). Use wavelet transform technology to remove baseline drift to ensure the stability of the signal. Then, calculate and subtract the average value of the signal to eliminate the influence of the DC component on the signal. Finally, for those points in the signal whose amplitude values ​​exceed the average amplitude by more than three times, they are regarded as abnormal points and directly eliminated to ensure the accuracy and reliability of subsequent analysis. Through this series of preprocessing steps, a purer and more accurate pulse wave signal can be obtained, providing a reliable data basis for subsequent blood pressure estimation.

[0035] Step 4: Extract the correlation features between the multi-spectrum and radar signals. When processing the video signals of the three channels of RGB, the absorption degree of blood to light of different wavelengths and the depth of light penetration into the skin are taken into consideration. For example, red light can penetrate deep into the skin, so the pulse wave signal of the red channel usually reflects the change of arterial blood volume. The green channel carries the volume change information of small arteries and veins, while the signal of the blue channel mostly reflects the mechanical movement of the surface of the skin.

[0036] At the same time, combined with the channel data of millimeter-wave radar, the correlation features between radar signals and multispectral signals can be further extracted. Radar signals can capture tiny surface movements of the human body, including chest displacement caused by heartbeats, and these signals have potential correlation with the pulse wave information in multispectral signals.

[0037] When extracting features, the contrast correlation features between RGB channels and between radar channels and multispectral channels are compared, such as PWHA (pulse wave area), systolic time (Tsys), systolic signal rise area (Ssys), time from end-systolic to dicrotic notch, etc. These features can reveal subtle changes in hemodynamics at different wavelengths, thereby providing valuable information for blood pressure estimation.

[0038] In addition, we also focus on the direct correlation features between signals, such as time difference and amplitude difference. Considering the propagation process of blood flowing from the heart to the artery and then flowing back to the heart through the veins during the cardiac cycle, the time delay and amplitude change of signals between different channels can be inferred. These direct correlation features are also important for blood pressure estimation.

[0039] The key to extracting the above features is to accurately locate the relevant feature points of the signal. Through precise signal processing technology, these key features can be identified and extracted to provide reliable data support for the subsequent blood pressure estimation model. The key to extracting the above features is to locate the relevant feature points of the signal. The specific feature extraction example is as follows:

[0040] The comparison and correlation features are taken as examples of the ratio of PWHA between the two RG channel signals and the difference in the systolic rising area of ​​the two RG channel signals. The PPG amplitude can be calculated by locating the maximum and minimum values ​​in a single cycle of the pulse wave, and the half-amplitude point can be located by the PPG amplitude. Then the PWHA can be obtained by calculating the time difference between the half-amplitude points. Finally, the ratio of PWHA between the two RG channel signals can be simply calculated to obtain the ratio of PWHA between the two RG channel signals. The systolic rising area of ​​the two RG channel signals can be obtained by the integral method. First, the trough and peak of the signal in the cycle are located by peak detection, and then the integral between the two points is calculated to obtain the rising area. Finally, the difference in the systolic rising area of ​​the two RG channel signals can be obtained by simply calculating the difference.

[0041] The direct correlation feature takes the time difference between the maximum slope point of the R channel and the end-diastolic point of the G channel as an example. The maximum slope point of the R channel can be located by the peak of the first-order derivative of the pulse wave signal, and the end-diastolic point of the G channel can be located by the trough of the pulse wave signal. The time difference feature between the RG signals can be obtained by directly calculating the time difference between the two points.

[0042] Step 5: Build a blood pressure estimation model and establish a blood pressure measurement system for actual measurement. Repeat the above five steps to extract the features of different samples and reference blood pressure, and use these sample features and reference blood pressure to train a blood pressure estimation regression model. Finally, build a blood pressure measurement system, where the input of the system is the human body video signal, and the output is the measured real-time blood pressure and blood pressure health status.

[0043] Step 6: Perform facial recognition on the portraits captured by the camera and correspond them to establish a personal blood pressure data set in the primary medical system to facilitate early warning and clinical auxiliary diagnosis and treatment.

[0044] This invention innovatively combines multispectral camera and millimeter wave radar technology, achieving an unprecedented breakthrough in the field of blood pressure measurement. Multispectral camera, with its significant advantages such as efficient and accurate data acquisition and analysis capabilities, high resolution, high spectral resolution, flexibility and adaptability, can obtain pulse wave signals in multiple optical bands and reveal the physical and chemical properties of the photographed object. Millimeter wave radar, with its small antenna aperture, narrow beam, high tracking and guidance accuracy, as well as excellent anti-interference and anti-stealth capabilities, provides another unique signal acquisition method.

[0045] This combination not only realizes the unlimited signal acquisition area, any skin area exposed in front of the camera lens can be used as a target for blood pressure measurement, but also has significant advantages in data processing and anti-interference. In practical applications, users can obtain blood pressure data in real time and accurately through this system without fixing their posture or restricting their movement. This blood pressure measurement method not only improves the accuracy and reliability of the measurement, but also greatly enhances the convenience and flexibility of the user experience, making blood pressure monitoring more intelligent and efficient.

Claims

1. A non-contact blood pressure measurement method combining visible light vision and millimeter wave radar, the method comprising: Step 1: Use a multi-spectral camera to collect human video signals and select the best region of interest for tracking, use a millimeter-wave radar to send a transmission signal to the pulse of the organism, collect echo data, and use a continuous electronic blood pressure meter to measure real-time continuous blood pressure; the multi-spectral camera resolution is at least 720p, the frame rate range is set to 15-30fps according to the pulse wave frequency, the collection site is not limited, and the best region of interest is the skin area in the video; Step 2: Convert the video signal into a time series signal of each channel; Step 3: Extract the millimeter wave radar IQ two-way signal; A signal with a large rate of change is selected for subsequent processing, and this time series signal is a type 2 signal; Step 4: Preprocess the class 1 signal and class 2 signal of each channel to remove non-pulse wave components. The specific operations include denoising, drift removal, DC component removal, and abnormal point removal. Step 5: Extract signal correlation features; The three channels of the class 1 signal and the one channel of the class 2 signal are combined into 4-channel data, and the half-width of the pulse wave and the rising area of ​​the heart systolic period of the 4-channel data are extracted. Then, the features extracted from each channel are subtracted and compared with each other to obtain the first type of signal correlation features; the systolic peaks, troughs and the maximum slope points of the pulse waves of each channel are located, and the time differences and amplitude differences of the above feature points between each channel are directly calculated to obtain the second type of signal correlation features; Step 6: Construct a blood pressure estimation model and establish a blood pressure measurement system; Repeat the above 6 steps to extract different sample features and obtain samples. A blood pressure estimation model is constructed based on all sample features and real-time blood pressure measurement values. The input of the blood pressure estimation model is the two types of signal correlation features extracted in step 5, and the output is the blood pressure value. Use the estimation model to establish and train a blood pressure measurement system. Step 7: Use the trained blood pressure measurement system to measure the blood pressure of the new target.

2. A non-contact blood pressure measurement method combining visible light vision and millimeter wave radar as claimed in claim 1, characterized in that: The specific method of step 2 is: Step 2.1: Divide the acquired image into three channels, namely, images of red spectral band, green spectral band and blue spectral band; Step 2.2: After channel separation, the average pixel grayscale value in the best region of interest of each channel image is calculated respectively, and the calculated value is used as the amplitude value of the corresponding channel signal at that moment. This time series signal is a class 1 signal.

3. The non-contact blood pressure measurement method combining visible light vision and millimeter wave radar as claimed in claim 1, characterized in that: The blood pressure estimation model constructed in step 6 is a multiple linear regression model, which uses a part of the features in step 5 and the real-time blood pressure values ​​to form a training set to obtain the regression parameters of the regression model.

4. The non-contact blood pressure measurement method combining visible light vision and millimeter wave radar as claimed in claim 2, characterized in that: The calculation method of the average pixel grayscale value in step 2.2 is: The average pixel gray value calculation formula is: Where n is the number of image frames, N is the total number of pixels, X(n) represents the average pixel grayscale value of the nth frame image, Sn represents the ROI corresponding to the nth frame image, and f(r,c) represents the pixel value at the pixel point (r,c).

Images