CN116269268A

CN116269268A - High-precision continuous blood pressure measuring device and method

Info

Publication number: CN116269268A
Application number: CN202211444026.6A
Authority: CN
Inventors: 王太宏; 马思清
Original assignee: Southern University of Science and Technology
Current assignee: Southern University of Science and Technology
Priority date: 2022-11-18
Filing date: 2022-11-18
Publication date: 2023-06-23

Abstract

The application provides a high-precision continuous blood pressure measuring device and a method, and relates to the technical field of blood pressure measurement. A multi-mode pulse sensor is placed at the superficial temporal artery to respectively sense the volume pulse wave of the superficial temporal artery and the contact pressure of the probe. And amplifying and demodulating the sensor original signal output through a phase-locked amplifying unit based on heterodyne oscillation technology to obtain a pulse wave signal with low noise and high time resolution. Waveform characteristics in the pulse wave signal and the probe pressure signal are related to the aortic blood pressure, and a blood pressure prediction model can be established by using the two data, so that accurate estimation of the blood pressure is realized. According to the scheme, various pulse wave signals are measured at the superficial temporal artery, a blood pressure prediction model is built, interference of the viscoelastic effect of the micro-artery at the traditional measuring site (such as a finger) on pulse wave transmission is avoided, and the obtained blood pressure estimated value is closer to the real blood pressure of the aorta.

Description

High-precision continuous blood pressure measuring device and method

Technical Field

The invention relates to the field of blood pressure measurement, in particular to a high-precision continuous blood pressure measurement device and method.

Background

Hypertension is one of the major threats to human health, but according to world health organization statistics, only less than half of patients are treated in time. The reason for the missing diagnosis of the hypertension is that the clinical blood pressure meter cannot acquire the real blood pressure condition of the patient. Clinical blood pressure meters can only intermittently sample because blood pressure fluctuates over time or environmental changes. Therefore, developing a high-precision continuous blood pressure measuring device which is not limited by time and environment is of great importance to improve early detection and intervention capability of hypertension.

Continuous blood pressure can be obtained by using some interference means, such as a flattening pressure probe method, a certain pressure is applied to the skin through the pressure probe, and the subcutaneous arterial vessel wall is flattened to obtain an approximate blood pressure value of the measuring position; the flow control method is to tighten the arm or finger by the binding belt with adjustable tightening force to block local blood flow, and calculate the blood pressure by the tightening force and the blood pressure waveform when blocked. The above techniques all cause discomfort to the human body, and continuous blood pressure measurement for a long time cannot be realized.

For non-interfering continuous blood pressure measurements, the current mainstream idea is to map it to blood pressure values using some readily available physiological signals, using mathematical models. These physiological signals are Electrocardiographic (ECG) signals, photoplethysmogram (PPG) signals, bioimpedance signals, and the like. Since PPG signals are most readily available, most continuous blood pressure schemes use PPG signals for blood pressure prediction.

The existing PPG signals are mainly collected at the finger, but the artery at the finger is very tiny, and the flowing resistance of blood in the artery is large. In addition, the finger artery has a small content of elastic fibers and a large distribution of smooth muscles, and the viscoelastic effect of the finger artery is not negligible, so that pulse waves in the finger artery are easily influenced by contraction of the smooth muscles to generate distortion. The above factors reduce the correlation between the aortic blood pressure and the finger arteriole pulse waveform, resulting in poor fitting accuracy of the blood pressure model.

Some PPG sensors collect pulse waves at the wrist, where the radial artery is thicker, and the viscoelastic effect on the pulse waves is less pronounced than in the finger arteries. However, when the wrist is far away from the heart and the wrist walks and moves, the arm swings greatly, so that the propagation of blood flow to the wrist is seriously affected, and a great deal of noise is generated in the wrist pulse wave. This factor limits the use of wrist sleeveless blood pressure measurement devices in motion.

Furthermore, any form of blood pressure measurement at the upper limb (finger, wrist, arm) is not effectively applicable to patients suffering from upper limb arterial disease, such as upper limb arteriosclerosis, upper limb arterial thrombosis, or upper limb amputees.

Disclosure of Invention

In order to solve the technical problems, the invention provides the high-precision continuous blood pressure measuring device and the high-precision continuous blood pressure measuring method, which are used for obtaining the blood pressure value of a user by analyzing the pulse waveform of the temporal artery of the head of the user, improving the portability of blood pressure monitoring, improving the accuracy of blood pressure estimation, avoiding discomfort caused by long-time measurement, and can be used for measuring the blood pressure of patients with pathological changes of the upper limbs and people with disabled upper limbs, and simultaneously meeting the blood pressure measuring requirements of the patients with pathological changes of the upper limbs.

The measuring method of the invention is realized on the basis of the high-precision continuous blood pressure measuring device, and the high-precision continuous blood pressure measuring device comprises: the combined probe module is fixed at the superficial temporal artery through a fixing device and is used for detecting pulse signals at the superficial temporal artery; the signal processing module is used for driving the combined probe module to work, amplifying pulse signals output by the combined probe module, removing interference noise and obtaining a multi-mode pulse feature set through a feature extraction algorithm; and the blood pressure calculation module is used for inputting a pre-established blood pressure estimation algorithm by utilizing the multi-mode pulse feature set to obtain a blood pressure estimation value.

The combined probe module comprises a photoplethysmography (PPG) pulse sensing unit and a pressure sensing unit, and the PPG pulse signal and the probe pressure signal at the superficial temporal artery are respectively acquired. Variations in contact pressure between the probe and the skin can affect the flow of blood in the subcutaneous artery, resulting in distortion of the received PPG pulse signal. The acquired probe pressure signal is used for compensating the influence of probe pressure change on the PPG pulse signal, so that the noise resistance of the device in motion measurement is enhanced.

The superficial temporal artery is one of the main arteries of the head and is located in front of the ear with a diameter of approximately 2.4mm. The superficial temporal artery has a larger diameter and less smooth muscle distribution than the arterioles at the finger, and has a higher elastic reservoir effect, which means that its volume is more susceptible to blood pressure fluctuations. Meanwhile, the superficial temporal artery is an end branch of the carotid artery, is closer to the heart, and is not easily interfered by non-blood pressure related factors in the process of propagation of pulse waves. Therefore, the PPG pulse waveform obtained at the superficial temporal artery has higher correlation with the aortic blood pressure waveform, so that the blood pressure estimation module can map the PPG pulse waveform to the aortic blood pressure more easily, the complexity of an estimation model is reduced, and the accuracy of blood pressure estimation is improved.

The fixing device is light in weight, convenient to carry and good in comfort, and can measure blood pressure anytime and anywhere after being worn by a user.

Unlike traditional flattening pressure probe blood pressure measuring method, the combined probe module of the invention does not press artery, and only needs tiny pressure to ensure that the probe is well attached to skin. The pressure sensing unit does not directly measure the pressure in the artery, but measures the contact pressure between the probe and the skin, and the contact pressure is small, so that the combined probe module can not cause discomfort when being worn for a long time, and continuous blood pressure measurement for a long time is realized. The PPG pulse sensing unit works based on a photoplethysmography (PPG) principle, comprises a light source and a photosensitive element, and selects a laser diode as the light source to ensure the signal quality of the emergent light of the light source. The photosensitive element uses a high-gain silicon photomultiplier (SiPM), and is characterized in that a microarray is formed by a plurality of Single Photon Avalanche Diodes (SPAD) working in a Geiger mode, each SPAD can trigger an avalanche breakdown process after receiving photons, a very large current is generated, and the current is converted into voltage pulses after being subjected to resistance, so that signal output is realized. Compared with a Photodiode (PD) and an Avalanche Photodiode (APD) of a traditional silicon-based photosensitive device, the SiPM has the characteristics of high gain and high sensitivity, and can detect weak optical signals which are not sensed by the traditional device.

The combination of the laser diode and the SiPM is characterized in that the device noise in measurement is effectively reduced, the low light detection capability of the PPG pulse sensing unit is improved, the measurement capability of high signal to noise ratio is realized in a wider bandwidth range, the detected PPG signal has less noise, the error in the model fitting process is reduced, and the final blood pressure estimation precision is improved.

The pressure sensing unit comprises, but is not limited to, a thin film capacitance pressure sensor and a thin film piezoresistance pressure sensor with flexible electrodes, outputs probe pressure signals, has the advantage of small volume, is easy to integrate in a miniaturized combined probe module, and realizes portable wearable blood pressure measurement.

The conventional amplification process amplifies signals and noise at the same time, and any amplification operation will reduce the signal-to-noise ratio of the signals without bandwidth limitation or filtering, so as to ensure the quality of the received signals, and the signal processing module includes a phase-locked amplification unit, and performs denoising processing while amplifying the original signals. The phase-locked amplifying unit multiplies the input signal with a reference signal, a process also known as heterodyne/homodyne detection, and then applies an adjustable low-pass filter to the result, by which the signal of the frequency of interest can be distinguished from other noise frequency components.

The specific functions of the lock-in amplifying unit in the invention comprise: providing a reference driving signal for driving the PPG pulse sensing unit to work; the original signals output by the PPG pulse sensing unit are subjected to phase-locked demodulation and filtering respectively, and amplitude variation signals and phase variation signals of the PPG pulse sensing unit relative to a reference driving signal are extracted; and outputting the amplitude variation signal or the phase variation signal as a pure PPG pulse signal, and combining the pure PPG pulse signal with the probe pressure signal to output the pure PPG pulse signal as a multi-mode pulse signal.

Specifically, for the PPG pulse sensing unit, the reference driving signal acts on the PPG pulse sensing unit after being amplified by the signal amplifying circuit, and the laser diode is driven to work in a linear region, so that pulse luminescence is realized. The laser passes through the skin of the human body, the pulse wave signal is received by the SiPM, and the original signal output by the SiPM is used for phase-locked demodulation.

The signal processing module further comprises a pulse wave feature extraction unit, wherein the pulse wave feature extraction unit performs feature extraction on the multi-mode pulse signals to generate a multi-mode pulse feature set.

The feature extraction will extract features of the clean PPG pulse signal, such as: slope characteristics, calculating slope values between two characteristic points of pulse waves and waveform derivatives thereof, and linear combination among different slope values; area characteristics, calculating the area of an area enclosed between two characteristic points of pulse waves and waveform derivatives thereof, and linear combination of different area values; intensity characteristics, intensity values corresponding to a pulse wave and a characteristic point of its waveform derivative, and linear combinations of different intensity values. Time characteristics, calculating the time difference between two characteristic points of the pulse wave and the waveform derivative thereof, and linear combination of different time differences.

The characteristic points are points with obvious morphological characteristics of pulse wave waveforms, such as waveform peak values, valley points, dicrotic wave points and the like.

The feature extraction unit may also extract features of the probe pressure signal including, but not limited to, average pressure, pressure variance, start-stop time pressure difference. The probe pressure characteristic and the PPG pulse signal characteristic are combined and output into a multi-mode pulse characteristic set.

The blood pressure calculation module comprises a blood pressure calibration unit, and the blood pressure calibration unit only works when the blood pressure model calibration is needed. When the calibration is performed, firstly, the combined probe module is used for collecting the training PPG pulse signals and the training probe pressure signals of the user in different time periods, and meanwhile, the sphygmomanometer is used for collecting the training standard blood pressure values of the user. And inputting the training PPG pulse signal and the training probe pressure signal into the signal processing module to obtain a training multi-mode pulse feature set. And the blood pressure calibration unit performs model fitting by utilizing the training multi-mode pulse feature set and the training standard blood pressure value, and finally obtains a blood pressure estimation model. The model fitting method includes, but is not limited to, one or a combination of the following: linear regression, machine learning, deep learning.

The blood pressure calculation module further comprises a blood pressure estimation unit, the multi-mode pulse feature set returned by the signal processing module is directly input into the blood pressure estimation unit during daily use, a calibrated blood pressure estimation model is stored in the blood pressure estimation unit, and a blood pressure estimation value is obtained through calculation.

Compared with the prior art, the invention has the beneficial effects that:

the invention describes a high-precision continuous blood pressure measuring device and a method, which are not dependent on an inflatable cuff, are comfortable to wear, can realize all-weather high-precision continuous blood pressure measurement, and have great significance for continuous monitoring and daily prevention of cardiovascular diseases.

The above-mentioned superficial temporal artery blood pressure sensing mechanism has an advantage in that the superficial temporal artery has a higher elastic reservoir effect and its volume is more susceptible to blood pressure fluctuations than the conventional scheme of acquiring pulse wave signals from the arterioles at the finger. Meanwhile, the superficial temporal artery is an end branch of the carotid artery, is closer to the heart, and is not easily interfered by non-blood pressure related factors in the process of propagation of pulse waves. The PPG pulse waveform obtained at the superficial temporal artery has higher correlation with the aortic blood pressure waveform, so that the blood pressure estimation module can map the PPG pulse waveform to the aortic blood pressure more easily, the complexity of an estimation model is reduced, and the accuracy of blood pressure estimation is improved.

Compared with the traditional flattening pressure probe blood pressure measurement method, the combined probe module provided by the invention has the advantages that arteries are not pressed, and only tiny pressure is needed to ensure good coupling between the photoelectric probe and skin, so that discomfort is not caused when the combined probe module is worn for a long time, and long-time continuous blood pressure measurement can be met. In addition, the conventional PPG sensor does not consider the influence of probe pressure on PPG measurement, and the contact pressure change between the probe and the skin can influence the flow of blood in subcutaneous arteries, so that the received PPG pulse waveform is distorted. The combined probe module provided by the invention can synchronously collect the pressure signal of the probe on the skin while collecting the PPG pulse signal, can compensate the influence of the pressure change of the probe on the PPG pulse wave, and enhances the noise resistance of the device in motion measurement.

The novel PPG pulse sensing unit provided by the invention adopts a structure of combining the laser diode and the SiPM, has the advantages of effectively reducing device noise in measurement, improving the low light detection capability of the PPG pulse sensing unit and having high signal to noise ratio measurement capability in a wider bandwidth range.

The phase lock demodulation method has the advantages that the phase lock amplifier is used for demodulating the pulse wave modulation signal, noise is restrained while the signal is amplified, accurate measurement of weak signals is achieved, and the problem that characteristic points are not easy to detect due to the fact that signal noise ratio of the pulse wave signal is low, influence of external noise is large, and signal burrs are too much is solved. The phase-locked demodulation is combined with the novel PPG pulse sensing unit, so that the final measured PPG signal has less noise, errors in the model fitting process can be reduced, and the final blood pressure estimation accuracy can be improved.

In addition, most of the existing blood pressure measurement means are developed on the upper limbs (fingers, wrists, arms), and are not suitable for patients suffering from upper limb arterial diseases, such as upper limb arteriosclerosis, upper limb arterial thrombosis, or upper limb amputees. The blood pressure measurement scheme provided by the invention can carry out accurate blood pressure measurement on the head, so that the daily or clinical blood pressure measurement requirement of patients with the diseases can be met. In order to make the above objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered limiting in scope, and that other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a block diagram of an overall system provided by the present invention;

fig. 2 is a schematic diagram of a combined probe module for superficial temporal artery according to the present invention; FIG. 3 is a schematic diagram of a phase-locked amplification scheme provided by the present invention;

Fig. 4 is a schematic diagram of a PPG signal according to the present invention;

FIG. 5 is a schematic diagram of a feature extraction scheme provided by the present invention;

FIG. 6 is a flow chart of a blood pressure calibration provided by the present invention;

FIG. 7 is a schematic diagram of a blood pressure estimation according to the present invention;

FIG. 8 is a schematic diagram of a blood pressure estimation effect according to the present invention;

fig. 9 is a schematic diagram of a superficial temporal artery blood pressure estimation performance provided by the present invention;

fig. 10 is a block diagram of a lock-in amplifying unit according to the present invention;

FIG. 11 is a schematic view of a probe arrangement position provided by the present invention;

FIG. 12 is a schematic diagram of a finger-sensing blood pressure estimation performance according to the present invention;

in the figure: 1. a combined probe module; 111. an upper housing; 112. a lower housing; 12. a PPG receiving electrode; 121. a SiPM photosensitive unit; 122. a first light guiding material; 13. a PPG emitter; 131. a laser diode; 132. a second light guiding material; 141. a PPG receiving extreme pressure sensor; 142. a PPG emission extreme pressure sensor; 2. elastic nylon bandage

Detailed Description

For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. The components of the embodiments of the present application, which are generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the present application, as provided in the accompanying drawings, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

In the description of the present application, it should be noted that, the terms "upper," "lower," "inner," "outer," and the like indicate an orientation or a positional relationship based on the orientation or the positional relationship shown in the drawings, or an orientation or a positional relationship conventionally put in use of the product of the application, merely for convenience of description and simplification of the description, and do not indicate or imply that the apparatus or element to be referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present application.

In the description of the present application, it should also be noted that, unless explicitly specified and limited otherwise, the terms "disposed," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the terms in this application will be understood by those of ordinary skill in the art in a specific context.

Some embodiments of the present application are described in detail below with reference to the accompanying drawings. The following embodiments and features of the embodiments may be combined with each other without conflict.

Example 1:

as shown in fig. 1, a high-precision continuous blood pressure measuring device is constructed in this embodiment in such a manner as to include a combination probe module, a signal processing module, and a blood pressure calculating module.

Wherein, this combination probe module includes: a PPG pulse sensing unit and a pressure sensing unit; the signal processing module includes: a lock-in amplifying unit and a feature extracting unit; the blood pressure calculation module includes: a blood pressure calibration unit and a blood pressure estimation unit;

the PPG pulse sensing unit is connected with the phase-locked amplifying unit; the phase-locked amplifying unit is connected with the feature extracting unit; the pressure sensing unit is connected with the feature extraction unit; the characteristic extraction unit is respectively connected with the blood pressure calibration unit and the blood pressure estimation unit; the blood pressure calibration unit is connected with the blood pressure estimation unit; the blood pressure estimation unit calculates and obtains an estimated blood pressure value by utilizing the blood pressure estimation model output by the blood pressure calibration unit and the multi-mode pulse feature set output by the feature extraction unit.

Optionally, the combined probe module is fixed at the superficial temporal artery through a fixing device and is used for acquiring multi-mode pulse wave signals at the position.

In this embodiment, the combined probe module is fixed at the left superficial temporal artery through the 3D-printed head-mounted fixing frame, and is attached to the skin above the superficial temporal artery, where the heartbeat pulse of the superficial temporal artery can be sensed.

The structure of the combined probe module is shown in fig. 2, and comprises an upper shell (111), a lower shell (112), a PPG receiving electrode (12), a PPG emitter (13), a PPG emitting extreme pressure sensor (141) and a PPG receiving extreme pressure sensor (142);

wherein the sub-portion of the PPG receiver (12) comprises: a SiPM photosensitive unit (121) and a first light guide material (122);

the subsection of the PPG emitter (13) comprises: a laser diode (131) and a second light guiding material (132).

Wherein the PPG receiving electrode (12) and the PPG emitter (13) are main components of a PPG pulse sensing unit in the combined probe module;

the PPG receiving extreme pressure sensor (141) and the PPG transmitting extreme pressure sensor (142) are main components of a pressure sensing unit in the combined probe module. The first light guide material and the second light guide material are both composed of polymethyl methacrylate.

The distance between the PPG receiving electrode and the PPG emitting electrode is 1.2cm, the PPG receiving electrode and the PPG emitting electrode are fixed on the lower shell in a mutual bonding mode, the periphery of the PPG receiving electrode and the PPG emitting electrode is not contacted with the upper shell, and the PPG receiving electrode and the PPG emitting electrode can be compressed freely in a certain axial range. The vertical pressure applied by the upper part to the emitter and receiver probes is completely transferred to the pressure sensor at the bottom, so that the force sensed by the pressure sensor is the contact pressure of the probes to the skin. The pressure sensor can be a FSR400 type film piezoresistive sensor, the outer diameter of the sensor is 7mm, the diameter of the sensing area is 5mm, and the sensitivity range is 0.2N-20N.

The PPG receiving extreme pressure sensor and the PPG transmitting extreme pressure sensor are respectively connected to a transmitting circuit, the transmitting circuit converts the resistance values of the PPG receiving extreme pressure sensor and the PPG transmitting extreme pressure sensor into voltage signals, the voltage signals linearly correspond to the pressure, and then the pressure voltage signals from the PPG receiving pole and the PPG emitting pole are averaged and output as probe pressure signals.

For the PPG receiving electrode, the SiPM photosensitive unit comprises an SiPM element, a PCB and a connecting wire; the SiPM element can be of the type Onsimi MICRORB-10035-MLP-TR, and is welded on a PCB (printed circuit board) with a necessary pre-filter circuit and a necessary line interface, and is connected with the outside through two shielding wires; the two shielding wires are respectively SiPM bias voltage lines, connected with an external-30V bias voltage source, and SiPM signal output lines, and connected with the phase-locked amplifying unit.

For the PPG pulse sensing unit emitter, the wavelength of the laser diode is 850nm, the rated power is 10mW, the light is scattered by a frosted surface polymethyl methacrylate light guide material (132) with the diameter of 2mm, and the spot size of the light which is injected into the skin of a human body is 3.14mm ² The power per unit area is small, and the human body cannot be damaged after long-time irradiation; the laser diode is connected to a reference signal generating circuit in the lock-in amplifying unit via a shield line, which emits a reference driving signal for driving the laser diode to emit light.

The PPG pulse sensing unit outputs a PPG pulse signal and forms a multi-mode pulse signal together with the probe pressure signal.

With continued reference to fig. 1, the signal processing module is configured to amplify, denoise, extract features, and the like, the multi-mode pulse signal output by the combined probe module. The phase-locked amplifying unit is used for amplifying and denoising the PPG pulse signal output by the PPG pulse sensing unit. The feature extraction unit extracts the features related to blood pressure in the pure PPG pulse signal which is subjected to phase locking and amplification and the probe pressure signal which is output by the pressure sensing unit.

The principle of the lock-in amplifying unit is shown in fig. 3, wherein the lock-in modulation process is described as follows:

the signal generating circuit generates a sine wave reference driving signal with the amplitude of 800mV and the direct current bias of 2.6V at 1.5MHz, and the sine wave reference driving signal is used for driving the laser diode in the PPG pulse sensing unit to emit light. The high-frequency sine pulse modulated light emitted by the laser diode is injected into the skin, and the scattering path and the absorption intensity of the modulated light are changed due to the periodical change of the diameter of a blood vessel along with pulse pulsation, so that the modulated light emitted out of the skin carries pulse wave information;

with continued reference to fig. 3, the phase-locked demodulation process of the phase-locked amplifying unit is described as follows:

The outgoing modulated light is received by SiPM, converted into an electric signal and transmitted to the input end of a phase-locked amplifying unit, the input signal is divided into two paths, multiplied by a reference driving signal and a 90-degree phase-shift signal thereof respectively, and then passed through a low-pass filter with a cut-off frequency of 25Hz to obtain an in-phase related component X and a quadrature related component Y of the input signal relative to the reference driving signal; after the polar coordinate conversion, the amplitude variation R and the phase variation θ of the input signal with respect to the reference driving signal can be obtained. The amplitude variation R after phase-locked demodulation is regarded as a real pulse wave signal, and after 7.2kHz sampling, the amplitude variation R obtains a digitized pure PPG pulse signal for subsequent processing.

In this embodiment, the main function of the lock-in amplifying unit can be realized by an HF2LI type lock-in amplifier manufactured by Zurich Instruments company. The model phase-locked amplifier is internally provided with a programmable signal generating circuit for realizing the phase-locked modulation process and a multichannel phase-locked demodulation circuit for realizing the phase-locked demodulation process.

The following derivation process will show how the phase lock amplification process described in this embodiment suppresses signal noise:

The pulse wave original signal output by the SiPM is taken as an input signal of the lock-in amplifying unit, and can be expressed as follows according to the difference of signal components:

V _PPG (t)＝S _r (t)·(K _Dc +k _m (t))+V _n0 +V _n1

wherein S is _r (t) is the reference drive signal of the modulated laser diode, the static attenuation component K _DC The dynamic attenuation component k corresponds to a constant attenuation value of near infrared light passing through skin tissue or vein layers _m (t) dynamic attenuation values corresponding to near-infrared light passing through the arterial blood vessel layer, can be considered to be effective pulse wave signals; low frequencyNoise component V _n0 High frequency noise component V corresponding to skin surface noise induced by factors such as muscle mechanical movement and respiration _n1 Corresponding device noise, including laser diode thermal noise and SiPM dark count noise, etc.

According to Fourier theory, the periodic signal can be developed into the sum of an infinite number of sine and cosine linear independent functions, and for the convenience of analysis, the driving signal S is referenced _r (t) is set to have a frequency f ₀ And amplitude V ₀ Corresponding to the fundamental frequency component of the excitation signal, the noise signal is also decomposed into fundamental frequency signals of K main frequency bands, which can further represent the probe output voltage as:

wherein the frequency of each noise fundamental frequency component is f _n，k Phase is theta _n，k Amplitude is V _n，k 。

Mixing the reference driving signal with the input signal to obtain mixed signals:

signal term 1 in the mixed signal is an effective signal term including only pulse wave related signals:

signal term 2 in the mixed signal is a noise signal term:

low-pass filtering with a cut-off frequency slightly higher than the bandwidth of the Yu Maibo wave signal is applied to the mixed signal, taking into account the variations of signal term 1 and signal term 2 after filtering, respectively. For signal item 1, a low frequency pulse wave signal component is obtained:

for signal term 2, when the excitation signal frequency f is referenced ₀ Far above the filter bandwidth, the noise term fundamental frequency is far above the low-pass filter cutoff frequency, so that the noise term will be eliminated after filtering.

In this embodiment, the feature extraction unit is configured to perform feature extraction on the phase-locked amplified pure PPG pulse signal and the probe pressure signal.

The step of extracting the characteristics of the pure PPG pulse signal comprises artifact elimination, reference point extraction and characteristic calculation. The signal artifacts are generated because the probe is relatively displaced from the skin contact surface, and the artifacts are difficult to avoid because the user cannot keep absolute rest during the measurement. Artifacts can cause the baseline component of the signal to drift over a wide range, affecting the accuracy of subsequent fiducial point extraction.

In this embodiment, a wavelet filter is used to filter out the low frequency baseline wander, the mother wavelet is Coiflets 5, whose scale function is similar to the pulse wave waveform, and the number of wavelet decomposition layers is set to 15, because for a signal sampled at 7.2kHz, the frequency of the 15 th layer approximation is about 0.11Hz, close to the dominant frequency of the baseline wander. During wavelet reconstruction, the layer 15 approximation component is completely removed as baseline drift, and the detail components of layers 1-8 are also completely discarded as high frequency noise. The effect after wavelet filtering is shown in fig. 4.

The pure PPG pulse signal after artifact elimination is input into a feature extraction unit, and the feature extraction unit extracts the following datum points for the input pulse wave: the pulse starting point, namely the starting time of a cardiac cycle, is expressed as a pulse waveform valley point; the pulse ending point, namely the ending time of one cardiac cycle, is expressed as the next pulse waveform valley point after the pulse starting point; the peak of the pulse, i.e. the ejection peak of one cardiac cycle, appears as the peak of the pulse waveform. And automatically searching for the valley point and the peak of the pulse waveform by an algorithm for searching for the local extremum, and after determining the valley point, intercepting data between two adjacent valley points to form a cardiac cycle waveform for subsequent analysis.

As shown in fig. 5, the heart cycle is divided according to the pulse start point and the pulse end point, and for one heart cycle, the following characteristics of the pulse wave are calculated:

pulse intensity (PA), amplitude difference between start point and peak; a cardiac Cycle (CP), a time difference between a start point and an end point; rise Area (RA), waveform area enclosed between start point and vertex; a Fall Area (FA), a waveform area enclosed between the vertex and the termination point; 10% rise time (RT 10), the time it takes for the waveform to rise to 10% pulse intensity; 33% rise time (RT 33), the time it takes for the waveform to rise to 33% pulse intensity; 50% rise time (RT 50), the time it takes for the waveform to rise to 50% pulse intensity; 75% rise time (RT 75), the time it takes for the waveform to rise to 75% pulse intensity; 90% rise time (RT 90), the time it takes for the waveform to rise to 90% pulse intensity; 10% fall time (FT 10), waveform falls to the difference between the 10% pulse intensity moment and the termination moment; 10% fall time (FT 10), waveform falls to the difference between the 10% pulse intensity moment and the termination moment; 33% fall time (FT 33), waveform falls to the difference between 33% pulse intensity time and termination time; 50% fall time (FT 50), waveform falls to the difference between the 50% pulse intensity moment and the termination moment; 75% fall time (FT 75), waveform falls to the difference between 75% pulse intensity time and termination time; 90% fall time (FT 90), the waveform falls to the difference between the 90% pulse intensity moment and the termination moment. The above features extracted in one cardiac cycle constitute a pulse wave feature set of one cardiac cycle.

The step of feature extraction of the pressure sensor signal includes segmentation and feature calculation. The segmentation process segments the pressure sensor signal according to the start and end times of a cardiac cycle, and for the pressure sensor signal of a cardiac cycle, the following features are calculated: average pressure (MP), average probe pressure over the cardiac cycle; a Start Pressure (SP), a probe pressure at the start of a cardiac cycle; pressure change (DP), the difference between the probe pressure at the end of the cardiac cycle and the probe pressure at the beginning of the cardiac cycle. The above features extracted over one cardiac cycle constitute a set of probe pressure features for one cardiac cycle.

The multi-modal pulse feature set for one cardiac cycle includes a pulse wave feature set and a probe pressure feature set for one cardiac cycle.

With continued reference to fig. 1, the blood pressure calculation module includes a blood pressure calibration unit and a blood pressure estimation unit. The blood pressure calibration unit only works when the blood pressure model calibration is needed, and performs model fitting by utilizing the multi-mode pulse feature set and the corresponding reference blood pressure value when the pulse wave is acquired, so that a blood pressure estimation model is finally obtained. The blood pressure estimation unit works in daily measurement, and inputs the multi-mode pulse feature set into the blood pressure estimation model to obtain a blood pressure estimation value.

The blood pressure calibration flow is shown in fig. 6, and includes:

and a data acquisition stage: pulse wave data and reference blood pressure for a plurality of cardiac cycles are acquired over a period of time, in this embodiment the duration of the data acquisition phase is 18 minutes.

Obtaining a pulse wave feature set through a PPG pulse sensing unit in the combined probe module and the signal processing module;

acquiring a probe pressure characteristic set through a pressure sensing unit in the combined probe module and the signal processing module;

the pulse wave feature set and the probe pressure feature set are combined to be called a multi-mode pulse feature set;

by wearing a reference sphygmomanometer, the sphygmomanometer performs calibration measurement at a frequency of once per minute to obtain a reference blood pressure, and the model of the reference sphygmomanometer can be subjected to Omron J753 type upper arm cuff sphygmomanometer characteristic data set construction stage:

the feature data set may include: all multi-modal pulse feature sets and their corresponding reference blood pressure values obtained throughout the calibration period. In particular, for the calibration periodThe time required for the sphygmomanometer to complete the measurement is about 30 seconds, during which about 40 pulse feature sets are generated, and a measured blood pressure value, which is a common reference for all pulse feature sets during the period. For the feature dataset, wherein one sample corresponds to one cardiac cycle, described by a plurality of attributes, the attributes being one feature, such as 'pulse intensity (PA)', to which the sample corresponds; the characteristic dataset may be represented as a matrix X, with the rows of X representing samples corresponding to a reference blood pressure value y _i The column of X represents the sample attribute, and X (i, j) is the value of the sample i on the attribute j.

Fitting model training:

this stage may include: and iterating model parameters through a certain optimization algorithm, performing model performance evaluation, and considering that the model can better predict blood pressure and iterate when the evaluation value is smaller than a preset threshold value. The model in this embodiment is a linear model, and can be expressed as:

y＝ω ^T X+b+∈

where y is the reference blood pressure value vector, X is the feature dataset, ω, b is the model parameter, and ε is the error term.

Model parameter optimization, which is used for searching optimal model parameters through continuous optimization process, and improving the fitting effect of the model and data. In the embodiment, model parameter optimization is performed by using a coordinate descent method, wherein the coordinate descent method is a non-gradient optimization algorithm, and in each iteration, one-bit search is performed along a coordinate direction at a current point to obtain a local minimum value of a function;

the model performance evaluation has the effect of evaluating the fitting degree of the model and the data through a certain evaluation method. In this embodiment, the linear model is solved by using Elastic network (Elastic Net) regression, the Elastic network is a depth improvement form of multiple linear regression, and the following evaluation functions are used to evaluate the performance of the model by combining the features of ridge regression and Lasso regression:

Where N is the number of samples, y _i Is the reference blood pressure x _i Is a sample. The super parameters lambda and alpha are penalty parameters of the model. When α=1, the model degenerates to Lasso regression, and when α=0, the model degenerates to ridge regression. Therefore, the elastic network regression combines the advantages of the ridge regression and the Lasso regression, so that the variable selection can be performed, and the group effect is good. In this embodiment, 10-fold cross validation is used to find the λ value that achieves the minimum mean square error, α is directly set to 0.5;

and stopping iteration after the obvious convergence of fitting errors is observed or the iteration number reaches 500 times through the iteration number and error convergence condition of model parameter optimization, and outputting the parameter omega and b at the moment as a blood pressure estimation model.

As shown in fig. 7, when the blood pressure calibration is completed, a daily blood pressure measurement can be performed using the obtained blood pressure estimation model. Obtaining a pulse wave feature set through a PPG pulse sensing unit in the combined probe module and the signal processing module; acquiring a probe pressure characteristic set through a pressure sensing unit in the combined probe module and the signal processing module; the pulse wave feature set and the probe pressure feature set are combined to be called a multi-mode pulse feature set; the multi-mode pulse feature set is directly input into a blood pressure estimation model, so that a blood pressure estimation value can be obtained.

The effect of blood pressure estimation achieved according to the method described in this example is shown in fig. 8, which shows the original blood pressure estimation results of 20 subjects, and the data of different subjects are separated by black dashed lines; the black solid line is a reference blood pressure value measured by the sphygmomanometer, the white solid line is a blood pressure estimated value, and the two blood pressure estimated values are very close to each other, so that a good measuring effect is obtained. Figure 9 shows statistics of mean blood pressure estimation performance for all subjects, assessed using the following four indicators:

1) Correlation coefficient:

2) Average ofError:

3) Standard deviation:

4) Average absolute error:

the performance of the high-precision continuous blood pressure measuring device described in the embodiment meets the clinical application standard of the blood pressure measuring device, which is provided by the American medical instrument detecting institute (AAMI) and has an error of less than or equal to + -5 mmHg and an error standard deviation of less than or equal to 8mmHg, whether for diastolic blood pressure or systolic blood pressure estimation.

Example 2:

example 1 was repeated with the following differences: the phase-locked amplifying unit is realized based on a Field Programmable Gate Array (FPGA), and the used FPGA model is Xilinx kinex 7. The possible implementation structure is as follows:

amplifying the signal of the PPG receiving electrode by using an Instrumentation Amplifier (IA), wherein the instrumentation amplifier is connected with an analog input end of an analog-to-digital converter (ADC), and a digital output end of the ADC is connected to the FPGA; and converting the digital reference driving waveform into an analog signal by using a digital-to-analog converter (DAC), wherein an analog output end of the DAC is connected with the PPG emitter, a digital input end of the DAC is connected with the FPGA, and the FPGA provides the digital reference driving waveform for the DAC. In addition, the FPGA is connected with a USB chip and further connected with a terminal, such as a PC, through a USB structure; as shown in fig. 10, the FPGA sends a set of digital reference signals to a 16-bit digital-to-analog converter (DAC) and then outputs as analog reference drive waveforms through a laser diode driving circuit for driving the laser diode to emit light in pulses.

The model of the 16-bit DAC is AD5662ARMZ-1REEL7, the final output analog reference driving waveform is 1.5MHz, the amplitude is 800mV, and the DC bias is 2.6V sine wave signal. The high-frequency sinusoidal pulse modulated light emitted by the laser diode is injected into the skin, and the scattering path and the absorption intensity of the modulated light in the blood vessel are changed due to the periodical change of the diameter of the blood vessel along with the pulse pulsation, so that the modulated light emitted out of the skin carries pulse wave information. The outgoing modulated light was received by the SiPM and converted to a voltage signal, which was input to a low noise Instrumentation Amplifier (IA) model AD8421, the noise spectral density at 1kHz was 3.5 nV/Hz, and the amplification gain was set to 40dB. The amplified signals are input into a high-speed analog-to-digital converter (ADC), and the digital waveforms output by the ADC are transmitted to the FPGA through an SPI serial port. The ADC model was AD9269-40, with 16-bit precision, and samples were taken at 40MHz frequency. Phase-locked demodulation logic in the FPGA is realized through programming.

Specifically, the waveform signal input by the ADC is divided into two paths, multiplied by a reference signal and a 90-degree phase-shift signal thereof respectively, and then passes through a low-pass filter with a cut-off frequency of 25Hz to obtain an in-phase correlation component X and a quadrature correlation component Y of the input signal relative to the reference driving signal; after polar coordinate conversion, the amplitude variation R and the phase variation theta of the input signal relative to the reference driving signal are obtained. The amplitude variation R after phase-locked demodulation is considered as a real pulse wave signal, which is down-sampled to 7.2kHz inside the FPGA, transmitted to a USB chip (FT 2232H) through a parallel interface, and then transmitted to a Personal Computer (PC) through USB for subsequent analysis.

Example 3:

example 1 was repeated with the following differences: the combination probe module may be secured over other superficial arteries using appropriate means,

including but not limited to the superficial temporal artery, radial artery, carotid artery, digital artery and capillaries thereof. In this embodiment, as shown in fig. 11, the combination probe module (21) is fixed to the thumb of the left hand by using an elastic nylon strap (22), and a multi-mode pulse wave signal of the finger is obtained. Based on the finger multi-modal pulse wave signals, blood pressure estimates were developed on 20 subjects. Fig. 12 shows statistics of average blood pressure estimation performance of all subjects, and the method of this example has a larger blood pressure estimation error than the superficial temporal artery blood pressure estimation performance shown in fig. 9. This result demonstrates that the superficial temporal artery sensing mode is more beneficial in improving the overall accuracy of blood pressure estimation.

It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners as well. The apparatus embodiments described above are merely illustrative, for example, flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, the functional modules in the embodiments of the present application may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.

The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

It will be evident to those skilled in the art that the present application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.

Claims

1. A high precision continuous blood pressure measurement device, comprising:

the combined probe module is fixed at the superficial temporal artery through a fixing device and is used for detecting pulse signals at the superficial temporal artery;

the signal processing module is used for driving the combined probe module to work, amplifying pulse signals output by the combined probe module, removing interference noise and obtaining a multi-mode pulse feature set through a feature extraction algorithm;

and the blood pressure calculation module is used for inputting a pre-established blood pressure estimation algorithm by utilizing the multi-mode pulse feature set to obtain a blood pressure estimation value.

2. The apparatus of claim 1, wherein the combined probe module comprises a photoplethysmography, PPG, pulse sensing unit and a pressure sensing unit, acquiring a PPG pulse signal and a probe pressure signal at the superficial temporal artery, respectively.

3. The device of claim 2, wherein the PPG pulse sensing unit comprises a light source and a photosensitive element, the light source is a near infrared band laser diode, the photosensitive element is a silicon-based photodetector, including a silicon photomultiplier SiPM, and the pressure sensing unit comprises a thin film capacitive pressure sensor and a thin film piezoresistive pressure sensor with flexible electrodes.

4. The apparatus of claim 2, wherein the signal processing module comprises a lock-in amplifying unit, and wherein the functions of the lock-in amplifying unit include:

providing a reference driving signal for driving the PPG pulse sensing unit to work;

performing phase-locked demodulation and filtering on the PPG pulse signal output by the PPG pulse sensing unit, and extracting an amplitude variation signal and a phase variation signal relative to a reference driving signal;

and outputting the amplitude variation signal or the phase variation signal as a pure PPG pulse signal, and combining the pure PPG pulse signal and the probe pressure signal to output as a multi-mode pulse signal.

5. The apparatus of claim 1, wherein the signal processing module further comprises a pulse wave feature extraction unit that performs feature extraction on the multi-modal pulse signals to generate the multi-modal pulse feature set.

6. The apparatus of claim 1, wherein the blood pressure calculation module includes a blood pressure calibration unit, and the blood pressure calibration unit is configured to perform model fitting using the multi-modal pulse feature set and a reference blood pressure value when in a calibration state to obtain a blood pressure estimation model.

7. The apparatus of claim 6, wherein the blood pressure calculation module further comprises a blood pressure estimation unit, the blood pressure estimation unit inputs the multi-modal pulse feature set into the blood pressure estimation model to obtain a blood pressure estimation value.

8. A method of high precision continuous blood pressure measurement, characterized in that the method is performed by the high precision continuous blood pressure measurement device according to any one of claims 1 to 7, the method comprising:

the combined probe module acquires PPG pulse signals and probe pressure signals at the superficial blood vessels of the user;

the signal processing module performs phase-locked demodulation on the PPG pulse signal of the user by using a phase-locked amplification principle to obtain a pure PPG pulse signal, and the pure PPG pulse signal and the probe pressure signal are combined to form a multi-mode pulse signal of the user;

Extracting characteristics of the acquired multi-mode pulse waveforms of the user, and generating the multi-mode pulse characteristic set;

and determining the blood pressure value of the user according to the multi-mode pulse feature set and a pre-established relation model of the multi-mode pulse feature set and the blood pressure value.

9. The method of claim 8, further comprising, prior to the step of determining the blood pressure value of the user from the multimodal pulse feature set and a pre-established relationship model of the multimodal pulse feature set to blood pressure values:

collecting training PPG pulse signals and training probe pressure signals of the user in different time periods by utilizing the combined probe module, and collecting training standard blood pressure values of the user by utilizing a sphygmomanometer;

performing phase-locked demodulation on the training PPG pulse signal of the user by using the signal processing module to obtain a pure training PPG pulse signal, and combining the pure training PPG pulse signal and the training probe pressure signal to form a training multi-mode pulse signal of the user;

performing feature extraction on the training multi-mode pulse signals of the user to obtain the training multi-mode pulse feature set, and combining the standard blood pressure value to construct the training data set;

Obtaining a relation model between the multi-mode pulse feature set and the standard blood pressure value by using the training data set and using a model fitting method, wherein the model fitting method comprises one or a combination of the following steps: linear regression, machine learning, deep learning.