A method for measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample

Abstract

The application discloses a kind of blood sample dynamic viscosity, storage modulus and equivalent load mass measurement method, belong to the field of blood coagulation analysis, by establishing the equivalent model of mechanical vibration system when blood viscoelastic sensor works, the dependent relationship of model parameter to the dynamic viscosity, storage modulus and equivalent load mass of sample is obtained;Establish the equivalent circuit of coil in sensor, according to equivalent circuit, the expression of the dynamic impedance module of coil is deduced, the model parameters are identified by using the frequency spectrum data of dynamic impedance module and through model parameter identification method;Multiple gradient standard viscosity liquid samples are used for experiment, and the calibration coefficient is obtained, the dynamic viscosity, storage modulus and equivalent load mass are calculated according to the model parameters and calibration coefficient identified, through the above steps, the dynamic viscoelasticity and equivalent load mass of sample in coagulation process can be measured simultaneously.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of blood coagulation analysis, and in particular to a method for measuring the dynamic viscosity, storage modulus and equivalent load mass of a blood sample. BACKGROUND

[0002] During the blood coagulation process, fibrin, platelets and blood cells form a three-dimensional cross-linked network structure, and under the action of fibrinolysin, fibrin is dissolved. During this process, the blood viscoelasticity changes. By detecting the changes in blood viscoelasticity during coagulation through a detection device, the coagulation process can be qualitatively or quantitatively analyzed, which can help doctors understand the coagulation function information of patients and make accurate diagnosis and treatment.

[0003] CN201921454553.9 discloses a thrombelastometry sensor, the structure of which is shown in Figure 1 , mainly comprising a magnetic cap 1, a permanent magnet 2, a coil 3, a probe base 4, a vibrating spring 5, a bottom plate 6 and a probe head 7. The permanent magnet 2 is fixed in the accommodating hole at the top of the magnetic cap 1, and a stable and constant magnetic field is formed in the air gap between the permanent magnet 2 and the magnetic cap 1. The probe base 4 is movably sleeved on the bottom of the permanent magnet 2 at the top, the coil 3 is single-directionally wound in the annular groove at the top of the probe base 4, and the coil 3 is in the stable and constant magnetic field. The joint of the coil 3 is led out. In operation, the coil 3 is connected with an alternating current, the coil 3 is subjected to an alternating force in the stable and constant magnetic field, and vibrates along the central symmetry axis direction, driving the probe base 4 to move synchronously. The vibrating spring 5 fixed with the probe base 4 provides a restoring force for the vibration, the probe head 7 is fixed on the probe base 4, and the lower end of the probe head 7 is inserted into the measured blood. The mechanical properties of the measured blood will affect the movement state of the moving part of the sensor.

[0004] Before blood coagulation, viscosity is dominant, and with the progress of coagulation, the role of elasticity gradually increases. Both viscosity and elasticity are important physical parameters reflecting coagulation function. Most of the existing detection technologies can only realize the detection of a single parameter of viscosity or elasticity. The existing representative blood viscoelasticity detection devices include the thrombelastography (TEG) of the Haemonetics company of the United States, the rotational thrombelastometry (ROTEM) of the Tem company of Germany and the platelet function analyzer (Sonoclot) of the Sienco company of the United States. The instruments such as TEG 5000, ROTEM delta and TEG 6s essentially detect the shear modulus of the whole blood sample, reflecting the elastic properties of the blood; while the Sonoclot coagulation and platelet function analyzer essentially detects the viscosity of the whole blood sample. They all do not realize the synchronous detection of viscosity and elasticity. The information of the coagulation system reflected by the elasticity and viscosity parameters of the blood is not completely the same, and it is necessary to simultaneously grasp the elasticity and viscosity information in the coagulation process. SUMMARY

[0005] In order to overcome the deficiencies of the prior art, one of the purposes of the present application is to provide a blood sample measurement method capable of simultaneously measuring dynamic viscosity (reflecting viscosity properties), storage modulus (reflecting elasticity properties) and equivalent load mass in the blood clotting process.

[0006] One of the purposes of the present application is achieved by adopting the following technical solutions:

[0007] A blood sample dynamic viscosity, storage modulus and equivalent load mass measurement method, comprising the following steps:

[0008] S1: Establishing an equivalent model of a mechanical vibration system when a blood viscoelastic sensor is working, wherein , wherein, is the elastic coefficient of the equivalent spring, is the total equivalent elastic coefficient of the sensor spring; is the storage modulus of the sample; is a constant determined by the geometry of the measuring cup and the probe; is the damping coefficient of the equivalent damper, is the total damping coefficient of the sensor spring; represents the radiation resistance of the spring when radiating the sound field; is the dynamic viscosity of the sample; is the mass of the equivalent mass block; represents the additional radiation mass of the sensor spring when radiating the sound field; represents the sum of the probe's own mass and the equivalent mass of the spring; is the equivalent load mass of the sample to be measured. The and are normalized with respect to the constant , is the electromechanical coupling coefficient of the blood viscoelastic sensor, and let ; from the above formula, we have: ,

[0009] , wherein the coefficients and are expressed as ;

[0010] S2: Establishing an equivalent circuit of the coil in the sensor, according to the equivalent circuit, the motional impedance of the coil is , is the angular frequency, is the imaginary unit; the modulus of , the normalized and are substituted into to obtain , according to the least square optimization model is established based on the spectrum, the initial value of the model parameter vector is searched and the corresponding weight of each experimental condition is determined, the nonlinear least square fitting is performed according to the parameter vector and the weight, and the parameters and are obtained.

[0011] S3: the experiment is performed on the standard viscosity liquid sample with multiple gradients, the frequency spectrum of the amplitude of the dynamic impedance corresponding to each sample is measured , the dynamic viscosity of the standard viscosity liquid sample is obtained according to step S2 , the dynamic viscosity is taken as the value of the dynamic viscosity , the linear regression result of the dynamic viscosity and the normalized damping coefficient is obtained, the slope of the linear regression equation is the calibration coefficient , the intercept of the linear regression equation is the calibration coefficient , the calibration coefficient is obtained through the experiment , and the calibration coefficient is also obtained through the experiment

[0012] S4: the parameters and obtained in step S2 and the calibration coefficients and obtained in step S3 are substituted into the expression of and , and the storage modulus , the dynamic viscosity and the equivalent load mass are calculated.

[0013] Further, in step S2, the least square optimization model is established based on the spectrum of , and the specific process is as follows:

[0014] In the single measurement process of the spectrum of , it is considered that the model parameters and are constant, the angular frequency is the experimental condition variable of the measurement , is used to represent the discrete experimental condition, and the discrete at is expressed as , and the relationship between and can be expressed as ,

[0015] In the formula, is the error term; model function is defined as ;

[0016] error term obeys a zero-mean Gaussian distribution with a standard deviation of , the model parameter vector is denoted as , according to the distribution characteristics of , the specific model parameter vector corresponding to is obtained is ;

[0017] for M experimental conditions , the joint probability density of all is the product of all , that is ; ;

[0018] The objective of least squares is to maximize , so that the model parameter vector is the minimum, that is, the least squares optimal solution, so the least squares optimization model can be represented as

[0019] ;

[0020] In the formula, is the weight, and the weight is determined by the standard deviation of the measured value corresponding to different experimental conditions .

[0021] Further, in step S2, the initial value of the model parameter vector is specifically:

[0022] In the working frequency range of the sensor, the influence of on the total impedance of the coil is small, so is ignored, and is obtained

[0023] In the formula, is the DC resistance of the coil wire, , ;

[0024] It can be represented by the maximum value of , that is, the following formula ;

[0025] When is ignored, the angular frequency​​​ Place, Take the maximum , using The maximum value of the spectrum and the angular frequency at this point, we can get And The value of,

[0026] Further calculation After, together with the following equations

[0027] ;

[0028] The solution of the equation set is the initial value of the model vector .

[0029] Further, the calculation Specifically:

[0030] Determine the impedance amplitude The maximum value And the angular frequency at this point , using the direct current resistance And , calculate ;

[0031] On the Curve, determine the frequency point ;

[0032] Calculate , .

[0033] Further, in step S2, determine the weight value corresponding to each experimental condition Specifically:

[0034] Use equal weight for least squares fitting;

[0035] Calculate the deviation of equal weight estimation;

[0036] Weight estimation.

[0037] Further, the least squares fitting using equal weight is specifically:

[0038] Let That is, assign equal weight to each measured value , with the model vector As the initial value of the nonlinear least squares fitting, the model parameter estimation value .

[0039] Further, the deviation of equal weight estimation is specifically:

[0040] The model parameter estimation value​ Substitute the model function , to obtain the estimated value of , calculate the deviation between the measured value and . .

[0041] Further, the weight estimation step is specifically:

[0042] Truncate using a rectangular window with a window width of , the standard deviation of the sequence obtained with a length of is , and the standard deviation is assigned to the measured variable at the center of the rectangular window, so that the weight at the center of the window is , and the center position of the rectangular window is changed in turn, so that it slides along the experimental condition variable axis, so that the weight corresponding to each is obtained. .

[0043] Further, in step S3, is the sensor electromechanical coupling coefficient, which is a constant, is the square of the electromechanical coupling coefficient, and the coefficient is calculated as follows:

[0044] Load a load with a mass of on the sensor probe, measure at least three different angular frequencies corresponding to the dynamic impedance amplitude data of the sensor coil, substitute at least three and the corresponding data into , and calculate and .

[0045] Load a load with a mass of on the sensor probe, is not equal to , measure at least three different angular frequencies corresponding to the dynamic impedance amplitude data of the sensor coil, substitute at least three and the corresponding data into , and calculate and by a model parameter identification method.

[0046] Substitute Mm1 , M m2 , M m_n1 as well as M m_n2 Substitution The coefficients were calculated. .

[0047] Furthermore, in step S3, the initial equivalent mass The total equivalent elastic coefficient of the two reeds of the sensor The results were obtained through experiments, specifically:

[0048] When there is no measurement sample, i.e., the sensor probe is only in contact with air, at least three different angular frequencies of the sensor coil are measured. Corresponding motional impedance amplitude The data will include at least three and the corresponding Data substitution The model parameter identification method is used to calculate and ;

[0049] Will and coefficients Substitution Calculations yielded and Since the probe is not in contact with the sample being measured at this time, therefore... They are respectively equal to and .

[0050] Compared to existing technologies, the method for measuring the dynamic viscosity, storage modulus, and equivalent load mass of blood samples in this invention establishes an equivalent model of the mechanical vibration system of a blood viscoelastic sensor during operation, thereby obtaining the dependence of model parameters on the dynamic viscosity, storage modulus, and equivalent load mass of the sample. It then establishes an equivalent circuit for the coil in the sensor, derives the expression for the motional impedance mode of the coil based on the equivalent circuit, and identifies the model parameters using the spectral data of the motional impedance mode through a model parameter identification method. and Experiments were conducted using standard viscosity liquid samples with multiple gradients to obtain calibration coefficients. The dynamic viscosity was then calculated based on the identified model parameters and the calibration coefficients. Energy storage modulus and equivalent load quality Through the above steps, the dynamic viscoelasticity and equivalent load mass of the sample during the coagulation process can be measured simultaneously. Attached Figure Description

[0051] Figure 1A perspective view of an existing thrombelastometry sensor in the background art;

[0052] Figure 2 A flow chart of the blood sample dynamic viscosity, storage modulus and equivalent load mass measurement method of the present application;

[0053] Figure 3 A schematic diagram of the equivalent model of the mechanical vibration system of the viscoelastic sensor in the present application;

[0054] Figure 4 An equivalent circuit diagram of the coil of the viscoelastic sensor in the present application;

[0055] Figure 5 A dynamic impedance amplitude-frequency characteristic diagram of the coil of the viscoelastic sensor in the present application;

[0056] Figure 6 A dynamic impedance phase-frequency characteristic diagram of the coil of the viscoelastic sensor in the present application;

[0057] Figure 7 A diagram of the actual measurement value and its equal weight value fitting of the dynamic impedance modulus;

[0058] Figure 8 A diagram of the standard deviation actual measurement value and estimated value of the dynamic impedance modulus;

[0059] Figure 9 A diagram of the nonlinear least squares fitting weight distribution result;

[0060] Figure 10 A dynamic viscosity calibration result;

[0061] Figure 11 A thrombelastogram signal diagram of two thrombelastometry quality control products;

[0062] Figure 12 A dynamic viscosity diagram of two thrombelastometry quality control products;

[0063] Figure 13 A storage modulus diagram of two thrombelastometry quality control products;

[0064] Figure 14 An equivalent load mass diagram of two thrombelastometry quality control products;

[0065] Figure 15 A thrombelastogram signal diagram of a whole blood sample;

[0066] Figure 16 A dynamic viscosity diagram of a whole blood sample;

[0067] Figure 17 A storage modulus diagram of a whole blood sample;

[0068] Figure 18Equivalent load mass chart for whole blood sample. DETAILED DESCRIPTION

[0069] The technical solutions in 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. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0070] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0071] Figure 2 Flow chart of the blood sample dynamic viscosity, elasticity and equivalent load mass measurement method of the present application, the blood sample dynamic viscosity, elasticity and equivalent load mass measurement method of the present application is specifically:

[0072] 1.1 Storage modulus and dynamic viscosity

[0073] Before coagulation, the viscosity of blood dominates, and the response to shear force is basically permanent deformation; after the blood starts to coagulate, the elastic effect gradually increases, and the deformation caused by shear force tends to be recoverable. The present application discusses the dynamic viscoelasticity of blood. Under the action of simple harmonic strain or stress, the mechanical response characteristics exhibited by the material are referred to as the dynamic viscoelasticity of the material. The elastic and viscous characteristics of the material can be obtained simultaneously by dynamic viscoelasticity detection. When dynamic viscoelasticity detection is performed, the flow field used is a small-amplitude simple harmonic shear flow field, which reflects the linear viscoelastic characteristics of the material being tested.

[0074] Let the small-amplitude simple harmonic strain acting on the blood sample be

[0075] ;

[0076] In the formula, is the angular frequency; t is time; is the imaginary unit; is the amplitude of the strain. If the blood sample is regarded as a linear body, the frequency of the simple harmonic stress and strain thereof is the same. The simple harmonic stress in the sample can be expressed as

[0077] ;

[0078] In the formula, the amplitude of the stress; the phase difference between the stress and the strain. Thus, may be further expressed as

[0079] ;

[0080] where, and are defined by the following two equations, respectively

[0081] ;

[0082] ;

[0083] G is the storage modulus of the sample, representing the contribution of elasticity; η is the dynamic viscosity of the sample, representing the contribution of viscosity.

[0084] 1.2 Equivalent model of the vibration system

[0085] When the probe of the sensor is immersed in a blood sample, the motion of the probe can be represented by a single-degree-of-freedom second-order damped equivalent model as shown in Fig. 1. The phasor of the velocity field of the sample in the measuring cup is Figure 3 s satisfies the passive wave equation as follows V

[0086] ;

[0087] where, is the Laplace operator; k is the complex wave number of the shear wave in the viscoelastic medium, defined by the following equation ;

[0088] In the formula, is the density of the sample under test.

[0089] The working frequency of the viscoelastic sensor ω is relatively low, the radial dimension of the sample under test can be considered to be much smaller than the wavelength of the shear wave in the sample, so that the wave equation of Eq. (1) satisfies the quasi-static approximation condition. Solving the wave equation of Eq. (1) under the quasi-static condition, we obtain the elastic coefficient of the equivalent spring and the damping coefficient of the equivalent damper Figure 3 are

[0090] ;

[0091] In the above two equations, and ​​​Ktotaleq and Ktotaldamp are the total equivalent elastic and damping coefficients of the two reeds of the vibrating viscoelastic sensor, respectively; Rtotrad is the total radiation resistance of the two reeds of the vibrating viscoelastic sensor when radiating the acoustic field; Ctot is a constant determined by the geometry of the measuring cup and the probe. Further, Figure 3 Mtot is the mass of the equivalent mass block M m which can be written as

[0092] ;

[0093] where, Mtotrad is the additional radiation mass of the two reeds of the vibrating viscoelastic sensor when radiating the acoustic field; Mtot is the sum of the mass of the probe itself and the equivalent mass of the whirling reed; Mload is the equivalent load mass of the sample under test.

[0094] According to Figure 3 , the equation of motion of the probe in phasor representation is

[0095] ;

[0096] where, V is the phasor of the probe velocity; F is the phasor of the driving force experienced by the probe, provided by the Lorentz force experienced by the coil, i.e. the following expression

[0097] ;

[0098] where, I c I is the phasor of the current in the coil; the electromechanical coupling coefficient of the sensor is defined by the following expression

[0099] (12);

[0100] where, L is a vector line element of a segment of the coil; B is the magnetic flux density vector; G is the closed loop geometry of the coil;

[0101] 1.3 The motional impedance of the coil

[0102] When a time-harmonic current is passed through the coil of the viscoelastic sensor, the resulting time-harmonic Lorentz force drives the coil, which produces a time-harmonic motion and, in turn, drives the probe of the sensor to produce a synchronous motion. The probe interacts with the sample under test, and its motion state is influenced by the mechanical properties of the sample. The coil and the probe are rigidly connected, so the phasor of the motion velocity of the coil in the magnetic field is also , the phasor of the induced electromotive force in the coil is for

[0103] ;

[0104] Changes in the induced electromotive force will lead to changes in the equivalent impedance of the coil. In summary, changes in the mechanical properties of the sample will modulate the equivalent impedance of the coil.

[0105] Since the sensor operates at a low frequency, the eddy current losses in the coil can be ignored. When considering the equivalent circuit of the coil, the equivalent resistance representing the eddy current losses in the high-frequency region can be neglected; only the DC resistance of the coil wire needs to be considered. coil inductance , and style The motional electromotive force represented The influence of small signals. Under the action of small signals, vibration sensors can be regarded as lumped parameter linear systems, so the equivalent circuit of the coil can be used. Figure 4 express. Figure 4 In, equivalent circuit elements and The values ​​are respectively

[0106] ;

[0107] ;

[0108] ;

[0109] in the formula Represents coefficients The square of.

[0110] Figure 4 middle, The component of the coil's equivalent impedance affected by the induced electromotive force is called motional impedance. The expression is as follows

[0111] ;

[0112] In the formula, the third term on the right is the motional impedance of the coil. ,

[0113] ;

[0114] The modulus and phase angle are respectively

[0115] ;

[0116] ;

[0117] Their frequency characteristics are respectively shown inFigure 5 and Figure 6 . The phase at the angular frequency is zero and its modulus takes a maximum value . The expression for

[0118] ;

[0119] The measured coil impedance minus the coil DC resistance and the inductive reactance is given by

[0120] .

[0121] 1.4 Dynamic viscoelasticity and equivalent load mass detection method based on dynamic impedance model parameter identification

[0122] Let and be normalized relative constants , let

[0123] ;

[0124] Substitute the above three equations into equation , then can be rewritten as

[0125] ;

[0126] From equation , it can be seen that the frequency spectrum of is uniquely determined by a specific set of model parameters and . In an ideal case, if the values of at three different angular frequencies are known, substituting equation (26) gives three equations, and solving this system of equations gives the uniquely determined model parameters and . However, the measured inevitably contains various errors, and it is difficult to ensure the accuracy of the model parameters determined from limited three-point data. In order to reduce the influence of random errors in the measurement data, it is necessary to establish a model parameter estimation result with better statistical performance on the basis of a large amount of measurement data. The number of linearly independent equations obtained in this way is much larger than the number of undetermined model parameters, i.e. it is an "overdetermined problem". Using the in a certain intervalBy using the spectral data to establish an overdetermined system of equations and identify the model parameters, relatively accurate model parameters can be obtained. , and .

[0127] Three normalized model parameters were obtained using the model parameter identification method. and Then, substituting into equations (7), (8), and (9) respectively, we can obtain...

[0128] ;

[0129] ;

[0130] ;

[0131] The correlation coefficients in the above three equations are determined by the following equations.

[0132] ;

[0133] ;

[0134] ;

[0135] .

[0136] 1.5 Parameter Identification of Motion Impedance Model Based on Least Squares Fitting

[0137] The model structure of motional impedance is known, and the identification of its model parameters falls under the category of parametric model identification problems in modern identification methods. This parametric model identification problem is solved by minimizing the error criterion function between the model and the measured data. Modern identification methods, in terms of their basic principles, can be divided into three types: least squares identification methods, gradient correction identification methods, and probability density approximation identification methods. The least squares method, among the least squares identification methods, is the most fundamental, its related theory is well-developed, and its application is the most widespread. The least squares method is adopted here.

[0138] 1.5.1 The model is established using the least squares method.

[0139] exist In a single measurement of the spectrum, the model parameters can be considered as... and Keep constant. Angular frequency ω For measurement The experimental condition variables, using Represents discrete experimental conditions. Discrete time Represented as Then The relationship between and can be expressed as

[0140] ;

[0141] where the model function is defined as

[0142] ;

[0143] The least square method is a special case of the more general maximum likelihood method. In applying the maximum likelihood method, the main concern is the probability of obtaining the measured values using the estimated model parameters, rather than the correct probability of the estimated model parameters themselves. Let the error term be subject to a Gaussian distribution with zero mean and standard deviation , and let the model parameter vector be denoted as . Then, according to the distribution characteristics of , the probability density of a specific model parameter vector is

[0144] ;

[0145] For M experimental conditions , the joint probability density of all is the product of all , i.e.

[0146] ;

[0147] The objective of the least square is to maximize . According to equation (37), the model parameter vector that minimizes is the optimal solution of the least square. Therefore, the least square optimization model can be expressed as

[0148] ;

[0149] where the weight is determined by the standard deviation of the measured values corresponding to different experimental conditions , i.e.

[0150] ;

[0151] The essence of the least square optimization model expressed by equation (38) is to find the model parameter vector ​to minimize the weighted sum of squares of residuals between the theoretical motional impedance amplitude determined therefrom and the measured values.

[0152] model function is a nonlinear function of the model parameters . To solve such a nonlinear least squares problem, one needs to start from some initial values of and use iterative methods including Gauss-Newton, gradient descent, Levenberg-Marquardt, etc. to find the global minimum point on the error surface

[0153] In applying the least squares method to solve specific engineering problems, the determination of the initial values of the fitting and the estimation of the weights have a great influence on the fitting results.

[0154] 1.5.2 Initial value estimation of model parameters

[0155] In using iterative methods to find the global minimum point on the error surface , it is particularly important to correctly estimate the initial values of the model parameter vector . A reasonable initial value not only avoids the iterative process converging to a local minimum point on the error surface, but also reduces the number of iterations and improves computational efficiency. The initial value of the model parameter vector should be as close as possible to the global minimum point. The following method is used for initial value estimation.

[0156] In the working frequency range of the sensor, the term in equation (17) has little effect on the total impedance . When determining the initial values of the model parameters, it is not necessary to perform strict and accurate solutions. Therefore, for equation (17), the term can be ignored and rewritten as

[0157] ;

[0158] In equation (40), the expression of is

[0159] ;

[0160] Let and substitute it into equation (40) to obtain

[0161] ;

[0162] According to equation (42), we have

[0163] ;

[0164] From equation (43), for any , if it satisfies​ then the following equation holds

[0165] ;

[0166] where , . Substituting equation (43) into equation (44) and using equation (45), we have ,

[0167] ;

[0168] Solving equation (45), we have

[0169] ;

[0170] In particular, taking , equation (46) can be simplified to

[0171] ;

[0172] When neglecting , can be represented by the maximum value of , i.e. the following equation

[0173] ;

[0174] Therefore can be calculated by the following equation

[0175] ;

[0176] According to the above analysis, on the basis of impedance analysis of the coil, measurement of its DC resistance and AC impedance, we can solve .

[0177] Step 1: solving Z mr , ω 0: i.e. determining the maximum value of impedance amplitude Z (j ω ) and the angular frequency Z mr at this point ω 0;

[0178] Step 2: solving r 0: using the DC resistance R e and Z mr , calculating r 0= Z mr / R ​​e ;

[0179] Step 3 solving ω 1、 ω 2: In Z (j ω ) curve, determine the frequency point Z (j ω 1) = 0 Z (j ω 2) = 0 r 0 1 / 2 R e ω 1、 ω 2;

[0180] Step 4 calculation Q ms : Q ms= ω 0 r 0 1 / 2 / | ω 2− ω 1|.

[0181] After the calculation of , and in combination with the previously obtained and , based on equation (21), equation (41) and equation (48), the following equation set can be solved simultaneously

[0182] ;

[0183] The solution of this equation set can be used as the initial value of the model vector .

[0184] 1.5.3 Weight estimation of the dynamic impedance amplitude measurement value

[0185] Before performing the nonlinear least square fitting of the model function , the corresponding weight of each experimental condition also needs to be determined. When the reliability of the measurement value is higher (i.e. the standard deviation is smaller), the weight assigned to it is higher, and vice versa. Since has non-flat frequency characteristics, the measurement of is usually an unequal precision measurement. Moreover, the time-varying nature of the blood sample mechanical properties during the coagulation process results in the time-varying nature of , and in a single sweep, it is impossible to obtain a sufficient number of samples and prior knowledge of the probability distribution of the measurement value under the same experimental condition.

[0186] For ​Unequal precision measurement, optimal weight Should be related to The variable. In a single sweep process, in order to shorten the measurement time, each angular frequency Only measured once . By The standard deviation of a single sample cannot directly derive the statistical information of And, when the blood coagulation measurement is carried out, because the model parameter vector Is a time-varying variable, so Can not be estimated in advance, and then can not be estimated from The prior knowledge of the probability distribution These factors increase the difficulty of weight estimation. In many applications, the weight is simply assigned to each measurement value At this time, a poor fitting result may be obtained.

[0187] Here, the weight estimation of the dynamic impedance amplitude measurement value is based on the following idea: in a relatively narrow frequency band, the sweep experiment conditions can be considered to be similar, so the standard deviation of In this frequency band should also be relatively close; if there are enough measurement data in this frequency band, the standard deviation at the center frequency can be represented by the standard deviation of all measurement data in this frequency band. The specific steps of this method are as follows:

[0188] (1) Least square fitting using equal weight

[0189] Let That is, assign equal weight to each measurement value Take the result of the initial value estimation method as the initial value, and perform least square fitting to obtain the model parameter estimate Although this parameter estimate may not be accurate enough, at least it can be considered that the deviation from the accurate value will not be too large.

[0190] (2) Calculate the deviation of equal weight estimation

[0191] First, substitute the model parameter estimate Obtained by using equal weight fitting in step (1) into equation (35) to obtain the estimate of The value of

[0192] ;

[0193] Then, calculate the deviation Between the measurement value And

[0194] ;

[0195] (3) Weight estimation​

[0196] When the sweep step is small, the requirement that the aforementioned experimental conditions are similar can be considered to be satisfied for a continuous segment of measurement values. A rectangular window with a center position at and a width of is used to truncate the sequence

[0197] ;

[0198] The standard deviation of the sequence of length is

[0199] ;

[0200] The standard deviation is assigned to the measurement variable at the center of the rectangular window, and thus the weight at the center of the window is

[0201] ;

[0202] By sequentially changing the center position of the rectangular window to slide along the experimental condition variable axis, the weight corresponding to each can be obtained.

[0203] A sensor is swept (the start and end frequencies are and , and the step of the experimental condition is ), and the weight estimation method described above is applied to the obtained measurement data, and the window width is 11 (as shown in Figure 7 ). The standard deviation estimation values at each are shown in Figure 8 . In order to facilitate comparison, the standard deviations of the measured data at several discrete angular frequencies are also shown. The measured values and the estimation values of the standard deviations at these angular frequencies are relatively consistent, thereby verifying the effectiveness of the weight estimation method. The corresponding estimated weights are shown in Figure 9 , and in the frequency band far from the natural frequency, the measurement accuracy is higher, and thus the weight is larger.

[0204] 1.5.4 Model parameter identification

[0205] After solving the two key problems of initial value estimation of model parameters and weight estimation, the weighted least squares fitting method can be used to identify the model parameters. Specifically as follows:

[0206] Step 1: According to the curve of impedance amplitude Z (j ω ), the natural frequency ω 0 and the maximum value of impedance amplitude Z ​mr and mechanical quality factor Q ms Solving the simultaneous equations to obtain the initial estimate of model parameters;

[0207] Step 2: Taking the initial estimate of Step 1 as the initial value, performing equal-weight fitting, and calculating the deviation of the equal-weight estimate;

[0208] Step 3: Applying the sliding section weight estimation method to the deviation of the equal-weight estimate to estimate the weight;

[0209] Step 4: Taking the result of the equal-weight fitting of Step 2 as the initial value, performing weighted fitting using the weight estimation result of Step 3 to obtain the final model parameter estimation result.

[0210] In Step 4, the initial value of the parameter for weighted fitting is not the initial estimate result of Step 1, but the initial value of Step 2, and the initial value of Step 1 is only used as the initial value of Step 2. The reason for this treatment is that experimental results show that the error of the equal-weight fitting result is smaller than the initial estimate error, especially when the frequency sweep step is large.

[0211] 1.6 Solution of the calibration equation coefficients

[0212] Storage modulus Dynamic viscosity and equivalent load mass are linearly related to the three normalized model parameters identified by model parameter identification and . To calibrate and , it is necessary to determine the coefficients , , and in equations (27), (28) and (29).

[0213] 1.6.1 Calibration equation of dynamic viscosity

[0214] Calibration of requires determining the coefficients and in equation (28). Standard viscosity liquids are used for calibration. In most conventional application scenarios, standard viscosity liquids can be regarded as Newtonian fluids.

[0215] First, take multiple gradient standard viscosity liquid samples for experiments, measure the frequency spectrum of the dynamic impedance amplitude corresponding to each sample , and apply the above dynamic impedance model parameter identification method to obtain For example, the parameters of the standard viscosity liquid used in the experiment are shown in Table 1. Experiments were performed on samples spl. 1-spl. 5. The standard viscosity liquid was kept at its constant temperature of 20°C by a temperature control device. After the temperature was stabilized for 5 minutes, the motional impedance spectrum of the voice coil was measured, and the model parameters were estimated. In order to reduce the influence of random errors in the sample addition amount on the calibration results, the experiments were repeated 5 times for spl. 1-spl. 5, and the estimation results of the model parameters are shown in Table 2.

[0216] Table 1 Parameters of the standard viscosity liquid used for calibration

[0217]

[0218] Table 2 Normalized damping coefficients obtained from 5 repeated experiments of samples spl. 1-spl. 5 R m_n

[0219]

[0220] Notes: R m_n The unit of

[0221] In the second step, the dynamic viscosity of the standard viscosity liquid sample is taken as the value of the dynamic viscosity in equation (28), and the normalized damping coefficient is taken as the average value of multiple experiments. Using the data of multiple standard viscosity liquid samples, linear fitting is performed, and the coefficient in equation (28) is obtained. For example, using the dynamic viscosity in Table 1 and the average value of 5 experiments in Table 2, the linear regression result of the dynamic viscosity Figure 10 and the normalized damping coefficient is shown in Table 3, and the corresponding calibration equation is (the dimensions of the quantities in the equation are in SI units)

[0222] ;

[0223] 1.6.2 Calibration equation of the elastic storage modulus

[0224] Since it is difficult to obtain standard substances similar to those used in dynamic viscosity calibration, and since the detection conditions are different, the detection results are also difficult to compare with those of third-party rheometers and other equipment. Therefore, it is difficult to calibrate the elastic storage modulus using standard substances or by comparison with third-party instruments. Here, the existing results of viscosity calibration, as well as the measured sensor Bl coefficient, are used to indirectly solve the coefficients of the elastic storage modulus calibration equation.​

[0225] First, examining equation (27) and equation (28), it can be found that, Calibration equation coefficient a The same as the coefficient of the first term of the viscosity calibration equation. a The value of has been obtained in the dynamic viscosity calibration.

[0226] Secondly, only need to determine . Using equation (30), equation (31), can be obtained

[0227] ;

[0228] Equivalent elastic coefficient of the reed , the coefficient of the sensor can be measured by experiment, and they and Bl are substituted into equation (57) to obtain a .

[0229] The calculation process of the coefficient is as follows:

[0230] Load a load with a mass of on the sensor probe, measure the dynamic impedance amplitude corresponding to at least three different angular frequencies of the sensor coil, substitute at least three and the corresponding data into , and calculate and ;

[0231] Load a load with a mass of on the sensor probe, not equal to , measure the dynamic impedance amplitude corresponding to at least three different angular frequencies of the sensor coil, substitute at least three and the corresponding data into , and calculate , and by the model parameter identification method;

[0232] Substitute M m1 , M m2 , M m_n1 and M m_n2 into , and calculate the coefficient​ .

[0233] Total equivalent spring constant of sensor The total equivalent spring constant of sensor is measured by experiment, specifically:

[0234] At least three different angular frequencies of sensor coil are measured when there is no measurement sample, i.e. the sensor probe only contacts air Corresponding motional impedance amplitude Data, at least three and corresponding Data are substituted into , and and are calculated by model parameter identification method;

[0235] Substitute and coefficient into , and is calculated Since the probe does not contact the measurement sample at this time, the at this time is equal to

[0236] For example, for the same sensor in the above dynamic viscosity calibration, the coefficient is 8.1465 Web·m -1 , and the equivalent spring constant of the vibrating reed is 522.606 kg·s -2 The calibration equation of the relative normalized equivalent spring constant of the sample storage modulus is obtained (Equations in the equation are in SI units)

[0237] ;

[0238] 1.6.3 Calibration equation of equivalent load mass

[0239] When the coefficient of the sensor has been measured, then the calibration equation of formula (29) about only needs to determine the initial equivalent mass of the vibration system . Similar to the equivalent spring constant of the vibrating reed and the Bl coefficient of the sensor, the initial equivalent mass can also be measured by experiment.

[0240] The initial equivalent mass is measured by experiment, specifically:

[0241] At least three different angular frequencies of sensor coil are measured when there is no measurement sample, i.e. the sensor probe only contacts air​ corresponding dynamic impedance amplitude data, at least three and corresponding data into , the model parameters are calculated by the model parameter identification method and ;

[0242] substitute and the coefficient into , the dynamic viscosity is calculated Since the probe does not contact the measured sample at this time, the dynamic viscosity at this time is equal to

[0243] For example, for the same sensor described above, the dynamic viscosity of 1.10611*10 -3 kg, the equivalent load mass of the sample relative to the normalized equivalent mass The calibration equation is

[0244] .

[0245] 1.7 Dynamic viscoelasticity and equivalent load mass detection results in the blood clotting process

[0246] Using the dynamic viscoelasticity and equivalent load mass detection method based on the dynamic impedance model parameter identification described above, two different thromboelastometry quality control products (Low-value QC and High-value QC, the main component is sodium citrate anticoagulant pig plasma) with different maximum amplitude levels of thromboelastogram are tested, and the results of Figures 11 to 14 are obtained, and the whole blood samples of healthy volunteers are tested, and the results of Figures 15 to 18 are obtained.

[0247] The present application discloses a method for measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample

[0248] By establishing an equivalent model of the mechanical vibration system of the blood viscoelasticity sensor in operation, the dependence of the model parameters on the dynamic viscosity, storage modulus and equivalent load mass of the sample is obtained; an equivalent circuit of the coil in the sensor is established, and the expression of the dynamic impedance modulus of the coil is derived according to the equivalent circuit; the model parameters and are identified by using the frequency spectrum data of the dynamic impedance modulus and through a model parameter identification method; a plurality of gradient standard viscosity liquid samples are used for experiments to obtain calibration coefficients, and the dynamic viscosity , storage modulus and equivalent load mass By the above steps, the dynamic viscoelasticity and the equivalent load mass of the sample in the blood coagulation process can be measured simultaneously, the application range of the blood viscoelasticity detection system is expanded, the current situation that most instruments can only detect a single parameter is changed, and the possibility of comprehensively evaluating the blood coagulation function according to more parameters is provided.

[0249] The above embodiments only express several implementation manners of the present application, the description is relatively specific and detailed, but it cannot be understood as the limitation of the patent scope of the present application. It should be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, which are equivalent modifications and evolutions of the above embodiments according to the essential technology of the present application, and these all belong to the protection scope of the present application.

Claims

1. A method of measuring the dynamic viscosity, the storage modulus and the equivalent load mass of a blood sample, characterized in that, The method comprises the following steps: S1: Establish the equivalent model of mechanical vibration system of blood viscoelastic sensor in working state , , , wherein, is the equivalent spring coefficient, is the total equivalent spring coefficient of sensor spring; is the storage modulus of sample; is a constant determined by the geometry of measuring cup and probe; is the equivalent damper coefficient; is the total damper coefficient of sensor spring; represents the radiation resistance of spring when radiating sound field; is the dynamic viscosity of sample; is the mass of equivalent mass block; represents the additional radiation mass of sensor spring when radiating sound field; represents the sum of probe self-mass and equivalent mass of spring; is the equivalent load mass of sample to be measured, and is a relative constant is normalized, is the electromechanical coupling coefficient of blood viscoelastic sensor, and ; from the above formula, , , , wherein, the expression of coefficient is ; S2: Establish the equivalent circuit of the coil in the sensor, and according to the equivalent circuit , the motional impedance of the coil , is the angular frequency, is the imaginary unit; the modulus of , the normalized and are substituted into to obtain , a least squares optimization model is established according to the frequency spectrum of , According to The least square optimization model of spectrum establishment is specifically: In The model parameters are assumed to be constant during a single measurement of the spectrum The angular frequency The experimental condition variable for the measurement is denoted by The discrete experimental condition at time is denoted by The discrete is denoted by The relation can be expressed as , wherein is the error term; the model function is defined as error term subject to a zero-mean Gaussian distribution with a standard deviation of Let the model parameter vector be denoted as Then, according to the distribution characteristics of , the specific model parameter vector corresponding to is For M each experimental condition , the joint probability density of all the variables is the product of all the individual probability densities, i.e. The goal of least squares is to minimize maximize the model parameter vector which is the least squares optimal solution, so the least squares optimization model can be expressed as wherein is a weight, the weights being different for different experimental conditions corresponding to the standard deviation of the measured values determined; Finding initial values of model parameter vectors and determining individual experimental conditions corresponding weights, The specific initial value of the model parameter vector is: In the operating frequency range of the sensor, The influence of the coil total impedance is small, so the term is neglected and we obtain , In the formulae, DC resistance of the coil conductor , ; may be represented by the maximum value of , i.e. the following ; neglect at the angular frequency , reaches a maximum , the value of the maximum of the spectrum and the angular frequency at this point, i.e. , Calculated Specifically: determining the impedance amplitude maximum value and the angular frequency here , using the direct current resistance , calculating ; In On the curve, determine the frequency point that makes ; Computing ; The calculation results are as follows Subsequently, the following equation group is solved simultaneously The solution of this system of equations is the initial value of the model vector x0. The parameter vector is obtained by performing a non-linear least squares fit according to the parameters and the weight values, and ; S3: using multiple gradient standard viscosity liquid samples to conduct experiments, measuring the frequency spectrum of the dynamic impedance amplitude of each sample , obtaining the dynamic viscosity of the standard viscosity liquid sample according to step S2 , taking the dynamic viscosity of the standard viscosity liquid sample as the dynamic viscosity value, obtaining the dynamic viscosity and the linear regression result of the normalized damping coefficient , the slope of the linear regression equation is the calibration coefficient , the intercept of the linear regression equation is the calibration coefficient , the calibration coefficient , the calibration coefficient Through experiments, it is found that , also measured by experiments; is the electromechanical coupling coefficient of the sensor, and is the square of the electromechanical coupling coefficient, and The calculation process of is as follows: A mass of is loaded on the sensor probe, and at least three different angular frequencies of the sensor coil are measured, corresponding dynamic impedance amplitudes Data are obtained, at least three and corresponding Data are substituted into , and and are calculated; loading a mass of on the sensor probe, is not equal to , at least three different angular frequencies of the sensor coil are measured corresponding motional impedance amplitudes data, at least three and corresponding data are substituted into and are calculated by a model parameter identification method Substitute M m1 , M m2 , M m_n1 and M m_n2 into , the coefficient is calculated; initial equivalent mass and sensor total equivalent spring constant measured experimentally, in particular: at least three different angular frequencies of the sensor coil are measured when the measurement sample, i.e. the sensor probe, is in contact with air only the corresponding motional impedance amplitude data, at least three and the corresponding data are inserted into by means of a model parameter identification method and ; Substitute and the coefficient into , and the calculation results are and Since the probe does not contact the measurement sample at this time, the value of at this time is equal to ; S4: Substitute the parameters obtained in step S2 and the scaling coefficients obtained in step S3 into the expression of step S1 and and to calculate the storage modulus , the dynamic viscosity and the equivalent load mass .

2. The method of measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample according to claim 1, characterized in that: In step S2, the respective experimental conditions are determined The corresponding weight values are specifically: The least square fitting is performed using equal weights; The bias of the equal weight estimation is calculated; The weight estimation.

3. The method of measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample according to claim 2, characterized in that: The least square fitting using equal weights is specifically: Let i.e. assign equal weight to each measurement with model vector Perform nonlinear least squares fit with initial values to obtain model parameter estimates 4. The method of measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample according to claim 3, characterized in that: The bias of the equal weight estimation is specifically: The model parameter estimate value is substituted into the model function , to obtain the estimate value of . The deviation between the measured value of is calculated.

5. The method of measuring dynamic viscosity, storage modulus and equivalent load mass of a blood sample according to claim 4, characterized in that: The weight estimation step is specifically: Using a window width of rectangular window right Cut off, the resulting length is The standard deviation of the sequence is This standard deviation is assigned to the measurement variable at the center of the rectangular window, thus obtaining the weight at the center of the window. for By sequentially changing the center position of the rectangular window and sliding it along the experimental condition variable axis, each... corresponding weights .

Images