Pulse wave signal preprocessing method and forehead physiological signal monitoring system

Abstract

The invention discloses a pulse wave signal preprocessing method and a forehead physiological signal monitoring system.The method comprises the steps that original pulse wave signals are obtained, noise of the original pulse wave signals is eliminated, time-frequency transformation is conducted on the original pulse wave signals through short-time Fourier transformation, and a first time-frequency graph is obtained; performing time-frequency rearrangement on the first time-frequency diagram according to a frequency axis to obtain a second time-frequency diagram; constructing a cost function based on the time-frequency energy and continuity constraint of the second time-frequency graph, and extracting a time-frequency ridge line by minimizing the cost function; based on the time-frequency ridge line, components used for reconstruction in the pulse wave signals are screened, and then signal reconstruction is carried out. According to the method, the signals are converted to the time-frequency domain with sparse distribution characteristics to be processed, screening and reconstruction of signal components are completed by adopting a simple rule, and the calculated amount is reduced while the signal quality is improved. The method is integrated into a forehead physiological signal monitoring system, so that the signal anti-interference capability is improved, and a physiological index calculation result is output in real time.

Description

Technical Field

[0001] The present invention belongs to the technical field of health monitoring systems, and in particular relates to a pulse wave signal preprocessing method and a forehead physiological signal monitoring system. Background Art

[0002] With the popularity of wearable medical devices, non-invasive physiological monitoring technology based on photoplethysmography (PPG) (such as heart rate, blood oxygen, and blood pressure detection) has become an important tool for health management. However, the original pulse wave signal is easily affected by multiple factors such as motion artifacts, environmental noise, and individual differences. In addition, ordinary users lack professional medical experience and the daily application scenarios are complex and changeable. The PPG signals collected by wearable devices are often mixed with a lot of noise and interference, which in turn damages the reliability of subsequent analysis. Therefore, in recent years, research related to PPG signal preprocessing has gradually become a hot topic. The mainstream technical routes include adaptive filtering, modal decomposition methods, sparse decomposition methods, deep learning methods, etc. In the process of realizing the present invention, the inventors found that there are at least the following problems in the prior art:

[0003] The implementation effect of methods such as adaptive filtering depends on the correlation between the reference signal and the noise. In current related research, the reference signal mainly comes from auxiliary sensors such as accelerometers, which increases the system hardware cost on the one hand, and cannot fully characterize the complex characteristics of interference components such as motion artifacts on the other hand; the computational complexity of methods such as sparse decomposition and deep learning is relatively high and cannot meet the real-time application requirements of products; the defects of modal decomposition methods are that on the one hand, methods represented by EMD (Empirical Mode Decomposition) and its improved versions lack theoretical basis, and on the other hand, due to the influence of factors such as modal aliasing and false components, the actual preprocessing effect is usually not ideal. Summary of the invention

[0004] Based on the problems existing in the prior art, the purpose of the embodiments of the present application is to provide a pulse wave signal preprocessing method and a forehead physiological signal monitoring system.

[0005] According to a first aspect of an embodiment of the present application, a pulse wave signal preprocessing method is provided, comprising:

[0006] Time-frequency transformation: obtaining an original pulse wave signal, eliminating noise from the original pulse wave signal and performing time-frequency transformation using short-time Fourier transform to obtain a first time-frequency graph;

[0007] Time-frequency post-processing: re-arranging the first time-frequency graph according to the frequency axis to obtain a second time-frequency graph;

[0008] Time-frequency ridge extraction: constructing a cost function based on the time-frequency energy and continuity constraint of the second time-frequency graph, and extracting the time-frequency ridge by minimizing the cost function;

[0009] Signal reconstruction: Based on the time-frequency ridge, the components used for reconstruction in the pulse wave signal are screened, and then signal reconstruction is performed.

[0010] Furthermore, in the time-frequency transformation step, a short-time Fourier transform represented by the following formula is performed:

[0011]

[0012] Where X(t,ω) is the first time-frequency diagram obtained after time-frequency transformation, s(t) is the input original pulse wave signal, w(t) is the window function, j is the imaginary unit, t is the time variable, ω is the frequency variable, and the window function adopts the Gaussian window shown in the following formula:

[0013]

[0014] where σ is the standard deviation parameter.

[0015] Furthermore, in the time-frequency post-processing step, the following formula is used to estimate the signal at a given time t and frequency ω ′ The instantaneous frequency is:

[0016]

[0017] The energy of the short-time Fourier transform is redistributed to the estimated instantaneous frequency position according to the following formula to make the energy more concentrated:

[0018]

[0019] Among them, T(t,ω) is the second time-frequency diagram obtained after post-processing, δ is the Dirac function, and θ is the frequency variable in the first time-frequency diagram before post-processing.

[0020] Furthermore, in the time-frequency ridge extraction step, the cost function C(t,ω) shown in the following formula is defined, which is composed of the time-frequency energy and continuity constraint of the signal:

[0021]

[0022] Where λ is the equilibrium parameter;

[0023] Using the dynamic programming algorithm, starting from time t = 0, by minimizing the global cost, we gradually search for the optimal path shown in the following formula along the time-frequency plane, that is, the time-frequency ridge line ω r (t):

[0024]

[0025] where γ(ω,ω ′ ) represents the smoothness penalty term.

[0026] Furthermore, the signal reconstruction step includes the following sub-steps:

[0027] Determine the fundamental wave candidate component set: based on the reasonable frequency range corresponding to the heart rate, set a screening threshold, and regard the components whose time-frequency points fall within the reasonable frequency range and whose proportion exceeds the screening threshold as fundamental wave candidate components and include them in the fundamental wave candidate component set;

[0028] For each fundamental candidate component C a (t,ω), calculate it and any other component C b The similarity between (t,ω);

[0029] For each fundamental candidate component C a (t,ω), select the first two components with the highest similarity as its harmonic candidate components C ah(i) (t,ω), i=1,2, based on the similarity between the fundamental candidate component and its harmonic candidate classification, calculate each fundamental candidate component C a Quality index of (t,ω);

[0030] The component with the highest quality index in the fundamental candidate component set is selected as the fundamental wave of the effective signal. The harmonic candidate component corresponding to this component is the harmonic of the effective signal. Then, the fundamental wave and the corresponding harmonic are reconstructed according to the following formula:

[0031]

[0032] Where real(·) means taking the real part of the signal, is the reconstructed signal, ω r (i, t) is the time-frequency ridge of the i-th component.

[0033] Furthermore, for each fundamental candidate component C a (t,ω), and calculate it and any other component C according to the following rules b The similarity between (t,ω):

[0034] Definition a (t),ω b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous frequency sequence corresponding to (t,ω) uses the coefficient of variation to measure the difference in the instantaneous frequency characteristics between the two components. The calculation formula is as follows:

[0035]

[0036] CVfreq (a, b) is the instantaneous frequency variation coefficient, ω a / b (t) = ω a (t) / ω b (t), Yes a / b (t), N is the length of the instantaneous frequency sequence;

[0037] Definition A a (t), a b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous spectrum amplitude sequence corresponding to (t,ω) uses the coefficient of variation index to measure the difference in the instantaneous spectrum amplitude characteristics between the two components. The calculation formula is as follows:

[0038]

[0039] CV amp (a, b) is the coefficient of variation of the instantaneous spectrum amplitude, A a / b (t) = A a (t) / A b (t), Yes A a / b The mean of (t), N is the data length;

[0040] Based on the instantaneous frequency variation coefficient and the instantaneous spectrum amplitude variation coefficient, the fundamental candidate component C is calculated using the Euclidean distance method. a (t,ω),C b The similarity R(a,b) between (t,ω) is as follows:

[0041]

[0042] According to a second aspect of an embodiment of the present application, there is provided a forehead physiological signal monitoring system, including a physiological signal acquisition device, a transmission network and a host computer;

[0043] The physiological signal acquisition device is used to acquire physiological signals, and the physiological signals include forehead temperature and pulse wave;

[0044] The transmission network is used to forward the physiological signal to a host computer;

[0045] The host computer preprocesses the physiological signal based on the method described in any one of claims 1 to 6, and calculates relevant health indicators based on the preprocessed physiological signal and dynamically outputs and displays them.

[0046] Furthermore, in the system:

[0047] The physiological signal acquisition device is built into the headband, including a pulse wave acquisition unit, a forehead temperature acquisition unit, and a microcontroller. The pulse wave acquisition unit includes an LED light source and a photoelectric sensor, and collects the original pulse wave signal by means of reflective PPG, and transmits it to the microcontroller after filtering, amplification, and analog-to-digital conversion; the forehead temperature acquisition unit uses a non-contact infrared measurement method to collect the forehead temperature signal, and transmits it to the microcontroller after filtering, amplification, and analog-to-digital conversion; the microcontroller is used to organize and encapsulate the received collected data into data packets and forward them to the transmission network;

[0048] The transmission network includes a host module and a slave module, the slave module is connected to the physiological signal acquisition device, and the host module is connected to the host computer, so as to forward the physiological signal from the signal acquisition device to the host computer;

[0049] The host computer receives the physiological signal data from the host module, unpacks and parses it, puts the parsed data into a data stream queue, performs signal preprocessing on it to suppress interference and noise, calculates heart rate and blood oxygen saturation indicators, and outputs the analysis results and the processed signal waveform to a UI interface to realize dynamic display of data;

[0050] Among them, the heart rate is calculated by the maximum peak value method of the amplitude spectrum:

[0051] Calculate the pulse wave signal amplitude spectrum as shown below:

[0052]

[0053] in is a pulse wave signal preprocessed by the method of any one of claims 1 to 6, S(ω) is the corresponding spectrum, real(·) indicates the real part, and imag(·) indicates the imaginary part;

[0054] S12: Find the maximum peak of the amplitude spectrum and its index frequency f within the frequency range corresponding to the reasonable heart rate range peak , then heart rate HR=f peak ×60;

[0055] The numerical calculation method of blood oxygen saturation is derived from Lambert-Beer's law, and the calculation formula is as follows:

[0056]

[0057] SpO2=aR 2 +bR+c

[0058] Where R is the perfusion index, SpO2 is the blood oxygen saturation, λ1 corresponds to the red light wavelength, λ2 corresponds to the infrared light wavelength, and I AC Represents the AC component in the signal, IDC It represents the DC component in the signal. The coefficients a, b, and c are determined by experimental calibration.

[0059] Furthermore, in the physiological signal acquisition device, the threshold for switching from the normal acquisition state to the stop acquisition state is set to V normal_limit , the threshold for switching from the stop acquisition state to the normal acquisition state is V stop_limit In the normal acquisition state, all LED light sources work normally. When the device is far away from the detected target, the received signal strength weakens to V normal_limit Next, turn off the red LED and reduce the infrared LED intensity to a low level to reduce power consumption. The physiological signal acquisition device switches from the normal acquisition state to the stop acquisition state. When the device approaches the detected target, the received signal strength increases to V stop_limit After the above, all LEDs resume normal operation, and the physiological signal acquisition device switches from the stop acquisition state to the normal acquisition state.

[0060] Furthermore, in the transmission network, the host module automatically identifies and connects to the slave module:

[0061] The working state of the host module is divided into a connection state and an idle state. When it is in the idle state, it periodically scans for connectable devices. When a matching slave is identified, it initiates and establishes a connection. The identification basis is the custom name of the target slave.

[0062] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0063] The pulse wave preprocessing method in this application is based on the hemodynamic analysis of the pulse wave, which is modeled as a harmonic model, and a signal preprocessing method based on time-frequency reconstruction is proposed. The main idea is to first analyze and obtain the time-frequency graph with high time-frequency aggregation, then extract the time-frequency ridges to achieve signal decomposition in the time-frequency domain, and finally customize the rules based on the harmonic characteristics to screen out the appropriate signal components and complete the signal reconstruction based on these components. In general, this method does not require auxiliary sensors, and chooses to convert the signal to the time-frequency domain with sparse distribution characteristics for processing based on a full theoretical analysis of the signal, and uses simple rules to complete the screening and reconstruction of signal components, which improves the signal quality while reducing the amount of calculation.

[0064] This application also proposes a forehead physiological signal monitoring system based on the above method, which integrates the above pulse wave signal preprocessing method and basic health index numerical calculation method in the host computer, thereby further improving the signal anti-interference ability at the software level and outputting the physiological index calculation results in real time, making it convenient for users to quickly understand the basic physical condition.

[0065] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0067] Figure 1 The figure is a flow chart of a pulse wave signal preprocessing method according to an exemplary embodiment.

[0068] Figure 2 It is a comparison diagram of the effects before and after time-frequency post-processing according to an exemplary embodiment.

[0069] Figure 3 is a diagram showing instantaneous frequency estimation results according to an exemplary embodiment.

[0070] Figure 4 is a comparison diagram of signal effects before and after signal preprocessing according to an exemplary embodiment.

[0071] Figure 5 It is a block diagram of a pulse wave signal preprocessing device according to an exemplary embodiment.

[0072] Figure 6 The diagram is a schematic diagram of an electronic device according to an exemplary embodiment.

[0073] Figure 7 The figure is an overall hardware block diagram of a forehead physiological signal monitoring system according to an exemplary embodiment.

[0074] Figure 8 It is an overall software block diagram of a forehead physiological signal monitoring system according to an exemplary embodiment.

[0075] Fig. 9 is a flowchart of physiological index calculation according to an exemplary embodiment. DETAILED DESCRIPTION

[0076] Here, exemplary embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0077] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0078] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0079] Figure 1 is a flow chart of a pulse wave signal preprocessing method according to an exemplary embodiment. Figure 1 As shown, the method may include the following steps:

[0080] (1) Time-frequency transformation: obtaining an original pulse wave signal, eliminating noise from the original pulse wave signal, and performing time-frequency transformation using short-time Fourier transform to obtain a first time-frequency graph;

[0081] For the original pulse wave signal to be processed, the effective frequency band of the pulse wave signal is first analyzed to be within 15HZ, so a low-pass filter is performed to eliminate out-of-band noise, and then the STFT represented by the following formula is used to realize time-frequency transformation:

[0082]

[0083] Where X(t,ω) is the first time-frequency diagram obtained after time-frequency transformation, s(t) is the input original pulse wave signal, w(t) is the window function, j is the imaginary unit, t is the time variable, ω is the frequency variable, and the Gaussian window with the best time-frequency aggregation shown in the following formula is used here:

[0084]

[0085] In this formula, σ is the standard deviation parameter, which is used to control the width of the Gaussian function. It determines the extent of the expansion of the window function w(t). The larger the value, the wider the window.

[0086] (2) time-frequency post-processing: re-arranging the first time-frequency graph according to the frequency axis to obtain a second time-frequency graph;

[0087] First, for any time t and frequency ω in the time-frequency domain, the true instantaneous frequency ω is re-estimated by the phase derivative ′ :

[0088]

[0089] Then, the energy of the time-frequency graph obtained in (1) is redistributed according to the following formula to further improve the time-frequency aggregation:

[0090]

[0091] Among them, T(t,ω) is the second time-frequency diagram obtained after post-processing, δ is the Dirac function, and θ is the frequency variable in the first time-frequency diagram before post-processing.

[0092] Comparison of the effects before and after post-processing Figure 2 As shown, (a) is the first time-frequency diagram, and (b) is the second time-frequency diagram. The details are partially magnified in the figure. It can be seen that the time-frequency aggregation is improved, and the energy of the time-frequency surface is concentrated on several ridges. Each ridge corresponds to a signal component, and the distances between some adjacent ridges are close, which has significant harmonic characteristics.

[0093] (3) Time-frequency ridge extraction: constructing a cost function based on the time-frequency energy and continuity constraint of the second time-frequency graph, and extracting the time-frequency ridge by minimizing the cost function;

[0094] First, define the cost function C(t,ω) shown in the following formula, which is composed of the time-frequency energy and continuity constraints of the signal:

[0095]

[0096] Where λ is a balance parameter that controls the importance of ridge smoothness.

[0097] Then, a dynamic programming algorithm is used to start from time t = 0 and gradually find the optimal path shown in the following formula along the time-frequency plane by minimizing the global cost, that is, the time-frequency ridge line ω r (t):

[0098]

[0099] where γ(ω,ω ′ ) represents the smoothness penalty term, which encourages the frequency changes of adjacent time steps to be small. The time-frequency ridges finally extracted are as follows: Figure 3 shown.

[0100] (4) Signal reconstruction: Based on the time-frequency ridge, the components in the pulse wave signal for reconstruction are screened, and then the signal is reconstructed;

[0101] First, the pulse wave is modeled as a harmonic model, and the modeling basis is as follows:

[0102] According to hemodynamic analysis, every time the human heart contracts, the pressure in the artery will fluctuate periodically. This fluctuation causes the oscillation of the arterial blood vessels, which is usually called the arterial pulse. The pulse wave propagates on the arterial wall and can propagate to the peripheral blood vessels. This pressure fluctuation propagating in space is the arterial pulse wave. The arterial pulse wave consists of two main parts: one is the forward wave generated by the contraction of the left ventricle, and the other is the reflected wave reflected from the end of the blood vessel. Therefore, the pulse wave can be simplified and modeled as a fundamental wave and its corresponding multiple harmonic components.

[0103] Here are the specific implementation steps:

[0104] S1: Determine the fundamental wave candidate component set: the reasonable frequency range corresponding to the heart rate is 0.5-3.3HZ, and the screening threshold is set to 60%. Then the components whose time-frequency points fall within this range and whose proportion exceeds the screening threshold are regarded as fundamental wave candidate components and included in the fundamental wave candidate component set;

[0105] S2: For each fundamental candidate component C a (t,ω), and calculate it and any other component C according to the following rules b The similarity between (t,ω):

[0106] S2.1: Define ω a (t),ω b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous frequency sequence corresponding to (t,ω) uses the coefficient of variation to measure the difference in the instantaneous frequency characteristics between the two components. The calculation formula is as follows:

[0107]

[0108] CV freq (a, b) is the instantaneous frequency variation coefficient, ω a / b (t) = ω a (t) / ω b (t), Yes a / b (t), N is the length of the instantaneous frequency sequence.

[0109] S2.2: Definition A a (t), A b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous spectrum amplitude sequence corresponding to (t,ω) uses the coefficient of variation index to measure the difference in the instantaneous spectrum amplitude characteristics between the two components. The calculation formula is as follows:

[0110]

[0111] CV amp (a, b) is the coefficient of variation of the instantaneous spectrum amplitude, A a / b (t) = A a (t) / A b (t), Yes A a / b (t), N is the data length.

[0112] S2.3: Based on the instantaneous frequency variation coefficient and the instantaneous spectrum amplitude variation coefficient, the fundamental candidate component C is calculated using the Euclidean distance method. a (t,ω),C b The similarity R(a,b) between (t,ω) is as follows:

[0113]

[0114] S3: For each fundamental candidate component C a (t,ω), select the first two components with the highest similarity as its harmonic candidate components C ah(i) (t,ω), i=1,2, based on the similarity between the fundamental candidate component and its harmonic candidate classification, calculate each fundamental candidate component C a Quality index of (t,ω):

[0115]

[0116] S4: Select the component with the highest quality index from the fundamental wave candidate component set as the fundamental wave of the effective signal. The harmonic candidate component corresponding to the component is the harmonic of the effective signal. Then, the fundamental wave and the corresponding harmonic are reconstructed according to the following formula:

[0117]

[0118] Where real(·) means taking the real part of the signal, is the reconstructed signal, ω r (i, t) is the time-frequency ridge of the i-th component, that is, the signal is reconstructed based on the video ridges of the three components of the fundamental wave and the corresponding harmonics.

[0119] At this point, the signal reconstruction process is completed. The effect comparison before and after reconstruction is as follows: Figure 4 As shown in the figure, the amplitude envelope of the reconstructed signal is more balanced and the periodic characteristics of the signal are more prominent, which is convenient for the detection of related parameters such as the heart rate in subsequent applications.

[0120] It can be seen from the above embodiments that the present application is based on the hemodynamic analysis of the pulse wave, and models it as a harmonic model, and accordingly proposes a signal preprocessing method based on time-frequency reconstruction. The main idea is to first analyze and obtain the time-frequency graph with high time-frequency aggregation, and then extract the time-frequency ridges to achieve signal decomposition in the time-frequency domain, and finally customize the rules based on the harmonic characteristics to screen out the appropriate signal components and complete the signal reconstruction based on these components. In general, this method does not require auxiliary sensors, and chooses to convert the signal to the time-frequency domain with sparse distribution characteristics for processing based on a full theoretical analysis of the signal, and uses simple rules to complete the screening and reconstruction of signal components, which improves the signal quality while reducing the amount of calculation.

[0121] Corresponding to the aforementioned embodiment of the pulse wave signal preprocessing method, the present application also provides an embodiment of a pulse wave signal preprocessing device.

[0122] Figure 5 FIG. 1 is a block diagram of a pulse wave signal preprocessing device according to an exemplary embodiment. Figure 5 , the device may include:

[0123] The time-frequency conversion module 21 is used to obtain the original pulse wave signal, eliminate noise from the original pulse wave signal and perform time-frequency conversion using short-time Fourier transform to obtain a first time-frequency graph;

[0124] A time-frequency post-processing module 22, configured to perform time-frequency rearrangement on the first time-frequency graph according to the frequency axis to obtain a second time-frequency graph;

[0125] A time-frequency ridge extraction module 23, configured to construct a cost function based on the time-frequency energy and continuity constraint of the second time-frequency graph, and extract the time-frequency ridge by minimizing the cost function;

[0126] The signal reconstruction module 24 is used to screen the components used for reconstruction in the pulse wave signal based on the time-frequency ridge line, and then perform signal reconstruction.

[0127] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0128] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0129] Accordingly, the present application also provides a computer program product, including a computer program / instruction, which implements the pulse wave signal preprocessing method as described above when executed by a processor.

[0130] Accordingly, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the pulse wave signal preprocessing method as described above. Figure 6 As shown in FIG. 1 , a hardware structure diagram of a pulse wave signal preprocessing device provided by an embodiment of the present invention is provided for any device having data processing capability, except Figure 6 In addition to the processor, memory and network interface shown, any device with data processing capability in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capability, which will not be described in detail.

[0131] Accordingly, the present application also provides a computer-readable storage medium on which computer instructions are stored, and when the instructions are executed by a processor, the pulse wave signal preprocessing method as described above is implemented. The computer-readable storage medium can be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card (Flash Card), etc. equipped on the device. Furthermore, the computer-readable storage medium can also include both an internal storage unit of any device with data processing capabilities and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and can also be used to temporarily store data that has been output or is to be output.

[0132] Based on the aforementioned embodiment of the pulse wave signal preprocessing method, the present application also provides a forehead physiological signal monitoring system, including a physiological signal acquisition device, a transmission network and a host computer, wherein the physiological signal acquisition device is used to collect physiological signals (including forehead temperature and pulse wave), and the transmission network (a BLE transmission network is used in one embodiment) is used to forward the physiological signals to the host computer, and the host computer preprocesses the physiological signals based on the aforementioned pulse wave signal preprocessing method, and calculates relevant health indicators based on the preprocessed physiological signals and dynamically outputs and displays them.

[0133] In a specific implementation, the physiological signal acquisition device is built into the headband for easy wearing, while performing low power consumption and miniaturization. The lower computer acquisition device implementation includes two parts: hardware design and software design. Figure 7 As shown, the hardware design of the device includes a pulse wave acquisition unit, a forehead temperature acquisition unit, and an MCU (Micro Controller Unit). The pulse wave acquisition unit includes an LED light source and a photoelectric sensor, that is, it collects the original pulse wave signal by means of reflective PPG, and then transmits it to the MCU after filtering, amplification, and analog-to-digital conversion. There are two original pulse wave signals in total, corresponding to the collection results under red light and infrared light respectively; the forehead temperature acquisition unit uses a non-contact infrared measurement method to collect forehead temperature signals, which are filtered, amplified, and transmitted to the MCU after analog-to-digital conversion; the MCU is used to organize and encapsulate the received collected data into data packets and forward them to the BLE transmission network. As shown Figure 8 As shown in the figure, the device software design realizes the acquisition of pulse wave and forehead temperature raw data streams through the I2C interface protocol, and then divides and assembles these raw data streams into data packets and forwards them to the BLE slave module through UART. The contents of the data packets are packet header identifier, data type identifier, raw data length, and raw data from the starting position. The software design of the physiological signal acquisition device also realizes the low-power operation logic of pulse wave detection, and sets the threshold value for switching from the normal acquisition state to the stop acquisition state as V normal_limit , the threshold for switching from the stop acquisition state to the normal acquisition state is V stop_limit In the normal acquisition state, all LED light sources work normally. When the device is far away from the detected target, the received signal strength weakens to V normal_limit Next, turn off the red LED and reduce the infrared LED intensity to a low level to reduce power consumption. The physiological signal acquisition device switches from the normal acquisition state to the stop acquisition state. When the device approaches the detected target, the received signal strength increases to V stop_limit After the above, all LEDs resume normal operation, and the physiological signal acquisition device switches from the stop acquisition state to the normal acquisition state.

[0134] The BLE transmission network implementation includes two parts: hardware design and software design. Figure 7 As shown, the hardware design of the BLE transmission network includes a slave module and a host module. The slave module is connected to the lower computer hardware through a soft wire, and the host module is connected to the PC device through a USB interface. Both use the UART interface protocol to communicate with the device end. Figure 3 As shown, the software design of the BLE transmission network realizes the automatic identification and connection of the BLE host to the target BLE slave and stable data transmission. Specifically, the working state of the BLE host is divided into a connection state and an idle state. When it is in an idle state, it periodically scans the connectable BLE devices. When a matching BLE slave is identified, it initiates and establishes a connection. The identification basis is the custom name of the target slave. When it is in a connected state, data can be transparently transmitted between the BLE master and slave. In order to ensure the transmission quality, the BLE connection interval, slave device delay, timeout time, transmission power and other parameters are optimized.

[0135] like Figure 8 As shown, the host computer software unpacks and parses the data after receiving it from the BLE host, and then puts the parsed data into the data stream queue, first performs signal preprocessing to suppress interference and noise, then calculates the heart rate and blood oxygen saturation indicators, and finally outputs the analysis results and the processed signal waveform to the UI interface to realize dynamic display of data, wherein the UI interface is implemented based on the QT framework.

[0136] The process of calculating the physiological indexes of heart rate and blood oxygen saturation in the host computer is as follows: Fig. 9 As shown in the figure, forehead temperature can be directly obtained by reading the raw data, while heart rate and blood oxygen saturation need to be numerically calculated based on the pre-processed pulse wave data.

[0137] The heart rate is calculated using the maximum peak value method of the amplitude spectrum. The specific steps are as follows:

[0138] S11: Calculate the pulse wave signal amplitude spectrum, as shown in the following formula:

[0139]

[0140] in is the preprocessed pulse wave signal, S(ω) is the corresponding spectrum, real(·) means taking the real part, and imag(·) means taking the imaginary part.

[0141] S12: The reasonable range of heart rate is 30-200 bpm, and the corresponding frequency range is 0.5-3.3 Hz. Find the maximum peak of the amplitude spectrum and its index frequency f within this range. peak , then heart rate HR=f peak ×60.

[0142] The numerical calculation method of blood oxygen saturation is derived from Lambert-Beer's law, and the calculation formula is as follows:

[0143]

[0144] SpO2=aR 2 +bR+c

[0145] Where R is the perfusion index, SpO2 is the blood oxygen saturation, λ1 corresponds to the red light wavelength, λ2 corresponds to the infrared light wavelength, and I AC Represents the AC component in the signal, I DC Represents the DC component in the signal. The coefficients a, b, and c are determined by experimental calibration and have different values ​​for different devices.

[0146] This system is designed based on the sports assistance scenario. In addition, it can also meet the daily health monitoring needs of various groups such as postoperative rehabilitation, chronic patients, and middle-aged and elderly people. The system equipment is selected to be placed on the forehead, which ensures the signal quality while minimizing the impact and inconvenience on the user; the system adopts a parity solution at the hardware level, and increases optimization at the software level to further reduce the overall cost. Specifically, this system includes acquisition equipment, preprocessing and numerical calculation algorithms, and supporting software development, which realizes the complete hardware and software link of signal acquisition, transmission, analysis, and output results. Taking the sports scene as the starting point, the user experience is optimized from multiple dimensions. First of all, on the acquisition device side, the design process adheres to the principles of miniaturization, lightweight, and low power consumption, and the device is embedded in the retractable headband, so that it is easy for users to wear; secondly, the device and the PC terminal adopt BLE wireless connection, which reduces power consumption while adjusting and optimizing the connection parameters to ensure the stability of connection transmission; the above-mentioned pulse wave signal preprocessing method and basic health index numerical calculation method are integrated in the PC terminal software, so as to further improve the signal anti-interference ability at the software level and output the physiological index calculation results in real time, so that users can quickly understand the basic physical condition.

[0147] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the contents disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.

[0148] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A pulse wave signal preprocessing method, characterized in that: include: Time-frequency transformation: obtaining an original pulse wave signal, eliminating noise from the original pulse wave signal and performing time-frequency transformation using short-time Fourier transform to obtain a first time-frequency graph; Time-frequency post-processing: re-arranging the first time-frequency graph according to the frequency axis to obtain a second time-frequency graph; Time-frequency ridge extraction: constructing a cost function based on the time-frequency energy and continuity constraint of the second time-frequency graph, and extracting the time-frequency ridge by minimizing the cost function; Signal reconstruction: Based on the time-frequency ridge, the components used for reconstruction in the pulse wave signal are screened, and then signal reconstruction is performed.

2. The method according to claim 1, characterized in that In the time-frequency transformation step, a short-time Fourier transform is performed as shown in the following formula: Where X(t,ω) is the first time-frequency diagram obtained after time-frequency transformation, s(t) is the input original pulse wave signal, w(t) is the window function, j is the imaginary unit, t is the time variable, ω is the frequency variable, and the window function adopts the Gaussian window shown in the following formula: where σ is the standard deviation parameter.

3. The method according to claim 1, characterized in that In the time-frequency post-processing step, the following formula is used to estimate the signal at a given time t and frequency ω ′ The instantaneous frequency is: The energy of the short-time Fourier transform is redistributed to the estimated instantaneous frequency position according to the following formula to make the energy more concentrated: Among them, T(t,ω) is the second time-frequency diagram obtained after post-processing, δ is the Dirac function, and θ is the frequency variable in the first time-frequency diagram before post-processing.

4. The method according to claim 1, characterized in that: In the time-frequency ridge extraction step, the cost function C(t,ω) shown in the following formula is defined, which is composed of the time-frequency energy and continuity constraint of the signal: Where λ is the equilibrium parameter; Using the dynamic programming algorithm, starting from time t = 0, by minimizing the global cost, we gradually search for the optimal path shown in the following formula along the time-frequency plane, that is, the time-frequency ridge line ω r (t): where γ(ω,ω ′ ) represents the smoothness penalty term.

5. The method according to claim 1, characterized in that The signal reconstruction step includes the following sub-steps: Determine the fundamental wave candidate component set: based on the reasonable frequency range corresponding to the heart rate, set a screening threshold, and regard the components whose time-frequency points fall within the reasonable frequency range and whose proportion exceeds the screening threshold as fundamental wave candidate components and include them in the fundamental wave candidate component set; For each fundamental candidate component C a (t,ω), calculate it and any other component C b The similarity between (t,ω); For each fundamental candidate component C a (t,ω), select the first two components with the highest similarity as their harmonic candidate components C ah(i) (t,ω), i=1,2, based on the similarity between the fundamental candidate component and its harmonic candidate classification, calculate each fundamental candidate component C a Quality index of (t,ω); The component with the highest quality index in the fundamental candidate component set is selected as the fundamental wave of the effective signal. The harmonic candidate component corresponding to this component is the harmonic of the effective signal. Then, the fundamental wave and the corresponding harmonic are reconstructed according to the following formula: Where real(·) means taking the real part of the signal, is the reconstructed signal, ω r (i, t) is the time-frequency ridge of the i-th component.

6. The method according to claim 5, characterized in that For each fundamental candidate component C a (t,ω), and calculate it and any other component C according to the following rules b The similarity between (t,ω): Definition a (t),ω b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous frequency sequence corresponding to (t,ω) uses the coefficient of variation to measure the difference in the instantaneous frequency characteristics between the two components. The calculation formula is as follows: CV freq (a, b) is the instantaneous frequency variation coefficient, ω a / b (t) = ω a (t) / ω b (t), Yes a / b (t), N is the length of the instantaneous frequency sequence; Definition A a (t), A b (t) are the fundamental candidate components C a (t,ω),C b The instantaneous spectrum amplitude sequence corresponding to (t,ω) uses the coefficient of variation index to measure the difference in the instantaneous spectrum amplitude characteristics between the two components. The calculation formula is as follows: CV amp (a, b) is the coefficient of variation of the instantaneous spectrum amplitude, A a / b (t) = A a (t) / A b (t), Yes A a / b The mean of (t), N is the data length; Based on the instantaneous frequency variation coefficient and the instantaneous spectrum amplitude variation coefficient, the fundamental candidate component C is calculated using the Euclidean distance method. a (t,ω),C b The similarity R(a,b) between (t,ω) is as follows:

7. A forehead physiological signal monitoring system, characterized in that: Including physiological signal acquisition equipment, transmission network and host computer; The physiological signal acquisition device is used to acquire physiological signals, and the physiological signals include forehead temperature and pulse wave; The transmission network is used to forward the physiological signal to a host computer; The host computer preprocesses the physiological signal based on the method described in any one of claims 1 to 6, and calculates relevant health indicators based on the preprocessed physiological signal and dynamically outputs and displays them.

8. The system according to claim 7, characterized in that In this system: The physiological signal acquisition device is built into the headband, including a pulse wave acquisition unit, a forehead temperature acquisition unit, and a microcontroller. The pulse wave acquisition unit includes an LED light source and a photoelectric sensor, and collects the original pulse wave signal by means of reflective PPG, and transmits it to the microcontroller after filtering, amplification, and analog-to-digital conversion; the forehead temperature acquisition unit uses a non-contact infrared measurement method to collect the forehead temperature signal, and transmits it to the microcontroller after filtering, amplification, and analog-to-digital conversion; the microcontroller is used to organize and encapsulate the received collected data into data packets and forward them to the transmission network; The transmission network includes a host module and a slave module, the slave module is connected to the physiological signal acquisition device, and the host module is connected to the host computer, so as to forward the physiological signal from the signal acquisition device to the host computer; The host computer receives the physiological signal data from the host module, unpacks and parses it, puts the parsed data into a data stream queue, performs signal preprocessing on it to suppress interference and noise, calculates heart rate and blood oxygen saturation indicators, and outputs the analysis results and the processed signal waveform to a UI interface to realize dynamic display of data; Among them, the heart rate is calculated by the maximum peak value method of the amplitude spectrum: Calculate the pulse wave signal amplitude spectrum as shown below: in is a pulse wave signal preprocessed by the method of any one of claims 1 to 6, S(ω) is the corresponding spectrum, real(·) indicates the real part, and imag(·) indicates the imaginary part; S12: Find the maximum peak of the amplitude spectrum and its index frequency f within the frequency range corresponding to the reasonable heart rate range peak , then heart rate HR=f peah ×60; The numerical calculation method of blood oxygen saturation is derived from Lambert-Beer's law, and the calculation formula is as follows: SpO2=aR 2 +bR+c Where R is the perfusion index, SpO2 is the blood oxygen saturation, λ1 corresponds to the red light wavelength, λ2 corresponds to the infrared light wavelength, and I AC Represents the AC component in the signal, I DC It represents the DC component in the signal. The coefficients a, b, and c are determined by experimental calibration.

9. The system according to claim 7, characterized in that In the physiological signal acquisition device, the threshold value for switching from the normal acquisition state to the stop acquisition state is set to V normal_limit , the threshold for switching from the stop acquisition state to the normal acquisition state is V stop_limit In the normal acquisition state, all LED light sources work normally. When the device is far away from the detected target, the received signal strength weakens to V normal_limit Next, turn off the red LED and reduce the infrared LED intensity to a low level to reduce power consumption. The physiological signal acquisition device switches from the normal acquisition state to the stop acquisition state. When the device approaches the detected target, the received signal strength increases to V stop_limit After the above, all LEDs resume normal operation, and the physiological signal acquisition device switches from the stop acquisition state to the normal acquisition state.

10. The system according to claim 7, characterized in that In the transmission network, the host module automatically identifies and connects to the slave module: The working state of the host module is divided into a connection state and an idle state. When it is in the idle state, it periodically scans for connectable devices. When a matching slave is identified, it initiates and establishes a connection. The identification basis is the custom name of the target slave.

Images