CN103479397B - Method for extracting color ultrasound parametric blood flow signal based on relaxation algorithm - Google Patents

Abstract

A color ultrasound parameterized blood flow signal extraction method based on a relaxation algorithm, the invention relates to a color ultrasound parameterized blood flow signal extraction method based on a relaxation algorithm. This method aims to solve the problem that the energy ratio of the clutter contained in the ultrasonic echo signal to the blood flow signal may be as high as 40dB because the scattering intensity of soft tissue is much greater than that of red blood cells, resulting in poor accuracy of blood flow signal extraction. The method uses a color ultrasonic detector to collect blood flow signals and transmits them to the receiving signal end; the receiving signal end establishes an affine model for the received ultrasonic echo signal, and uses a relaxation algorithm to estimate the parameters of the affine model to obtain the blood flow signal frequency and amplitude. The invention is applicable to the technical field of ultrasonic color blood flow imaging, and is applied to the research of various cardiovascular diseases.

Description

Color Ultrasound Parameterized Blood Flow Signal Extraction Method Based on Relaxation Algorithm

technical field

The invention relates to the technical field of ultrasonic color blood flow imaging.

Background technique

As a non-invasive blood flow imaging method, ultrasound color flow imaging is widely used in the research of various cardiovascular diseases. The clutter signals produced by the slow movement of the vessel wall and its surrounding soft tissue will have a great impact on the accuracy of the blood flow to be displayed. Failure to provide accurate blood flow information will increase the risk of misdiagnosis and the difficulty of evaluating the accuracy of long-term medical testing of patients. In practical applications, the accuracy of blood flow velocity estimation decreases for the following two reasons:

First, because the scattering intensity of soft tissue is much greater than that of red blood cells, the energy ratio of the clutter contained in the ultrasound echo signal to the blood flow signal may be as high as 40dB; this is because the clutter filter is often unable to adequately filter the clutter or Part of the blood flow signal is incorrectly filtered out, resulting in an error in blood flow estimation. In particular, even if the clutter can be sufficiently suppressed in the filtering stage, the white noise remaining in the signal after filtering will still cause errors in the mean and variance estimation of the Doppler frequency of the blood flow signal;

Second, under the real-time requirement of ultrasound imaging, the number of received Doppler signal pulses is limited, often lasting only 8-16 pulse times.

Contents of the invention

In order to solve the problem that the scattering intensity of soft tissue is much greater than that of red blood cells, the energy ratio of the clutter contained in the ultrasonic echo signal to the blood flow signal may be as high as 40dB, resulting in poor accuracy of blood flow signal extraction, thus proposing Color ultrasound parametric blood flow signal extraction method based on relaxation algorithm.

The color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm is as follows: the blood flow signal is collected by a color ultrasound detector, the receiving signal end establishes an affine model for the received ultrasound echo signal, and the relaxation algorithm is used to estimate the affine model parameters to obtain the frequency and amplitude of the blood flow signal.

The specific process of establishing an affine model for the received ultrasonic echo signal at the receiving end is as follows:

Use the received ultrasonic echo signal x as an affine model:

x≈Pα (1)

Among them, α is the amplitude strength vector,

P is the feature matrix:

P=[p(f ₁ ) p(f ₂ ) …p(f _k )…p(f _K )] (2),

The expression of the ultrasonic echo signal x:

x=[x(1) x(2) ... x(n) ... x(N)] ^T (3)

Among them, k is the number of main components, k=1, 2, ..., K, k and K are all positive integers; x(1) represents the first sampling value, x(2) represents the second sampling value , x(n) represents the nth sampling value, x(N) represents the Nth sampling value, n=1,2...,N, n and N are natural numbers; p(f _k ) represents the frequency vector of the kth principal component.

Using the relaxation algorithm to estimate the parameters of the affine model, the specific process of obtaining the frequency and amplitude of the blood flow signal is as follows:

Step A, estimate the frequency and amplitude of the first principal component, subtract the principal component from the input signal, estimate the frequency and amplitude of the second principal component, and perform step B;

Step B. Reversely update the frequency and amplitude of the first principal component according to the frequency and amplitude of the second principal component;

Step C, judging whether the frequencies of the two principal components are stable, if so, execute step D, if not, execute step A;

Step D. Estimate the frequency and amplitude of the third principal component according to the two principal components whose frequencies are stabilized, and update the frequency and amplitude of the third principal component according to the frequency and amplitude of the first principal component and the frequency and amplitude of the third principal component. The frequency and amplitude of the two principal components, or update the frequency and amplitude of the first principal component based on the frequency and amplitude of the second principal component and the frequency and amplitude of the third principal component;

Step E, combine the first principal component, the second principal component and the third principal component to obtain the energy of the combined signal, and judge whether the difference between the energy of the combined signal and the energy of the input signal is greater than or equal to greater than 1- δ, δ=10 ^-2 , if yes, select the Doppler frequency with the largest three principal component frequencies as the Doppler frequency of the blood flow signal, if not, perform step A.

The specific process of estimating the frequency and amplitude of the first principal component described in step A and subtracting the principal component from the input signal to estimate the frequency and amplitude of the second principal component is:

Step 1. Initialize the ultrasonic echo signal x, the ultrasonic echo signal contains K principal components, set K=1, substitute the ultrasonic echo signal x into formulas (9) and (10) to obtain the frequency of the first characteristic component estimated value of and an estimate of the amplitude of the first eigencomponent Execute step two;

$\begin{array}{cc}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ]</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span>$

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>$

Among them, f is a variable, the value range of this variable is [-fs/2, fs/2], fs is the sampling frequency, and the superscript H indicates the conjugate transpose of the vector;

Step 2. Let K ₀ =K ₀ +1, according to the estimated value of the frequency of the first feature component obtained in Step 1 and an estimate of the amplitude of the first eigencomponent Obtain x ₁ according to formula (11),

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>$

Use formulas (12) and (13) to obtain the estimated value of the frequency of the second characteristic component and an estimate of the amplitude of the 2nd eigencomponent According to estimates and Calculate x ₂ using formula (14),

$\begin{array}{cc}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& \mathrm{<span\; class="notranslate">\; mi</span>}{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ]</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 12</span>}<span\; class="notranslate">\; )</span>$

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 13</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; .</span>$

The specific process of updating the frequency and amplitude of the first principal component in reverse according to the frequency and amplitude of the second principal component described in step B is: according to the obtained _x2 , it is obtained by formulas (9) and (10) and

The specific process of judging whether the frequencies of the two principal components described in step C are stable is: the estimated value with estimates Make a difference, and judge whether the absolute value of the obtained difference is less than or equal to 10 ^-3 , if less than or equal to, the frequency of the two principal components is stable, and if greater, the frequency of the two principal components is unstable.

The present invention is different from the non-parametric blood flow signal extraction method, and the parameterized method does not need to design a clutter filter, but directly extracts the Doppler frequency of the blood flow signal from the ultrasonic echo signal. The present invention establishes an affine model for the input signal, and uses the relaxation algorithm to estimate the model parameters, thereby improving the estimated mean value and estimated variance of the Doppler frequency of the blood flow signal, thereby achieving the goal of improving the accuracy of blood flow signal extraction Objective: Compared with the popular parametric method (multi-signal classification), the accuracy can be increased by 40% when the energy ratio of clutter and blood flow signal is 40dB.

Description of drawings

Fig. 1 is the flow chart of the method for estimating affine model parameters by relaxation algorithm;

Figure 2 is a comparison curve between the relaxation algorithm and the parametric method for estimating the Doppler frequency of the blood flow signal when the C/S is 40dB; C represents clutter; S represents the blood flow signal Signal,

Fig. 3 is a comparison curve between the relaxation algorithm and the parameterized method for estimating the Doppler frequency of the blood flow signal when the C/S is 20dB;

Fig. 4 is a comparison curve when the C/S is 40dB for the variance of the estimation result between the relaxation algorithm and the parametric method;

Fig. 5 is a graph comparing the variance of the estimation result between the relaxation algorithm and the parametric method when the C/S is 20dB.

Detailed ways

Specific Embodiment 1. The color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in this embodiment is as follows: a color ultrasound detector is used to collect the blood flow signal, and the receiving signal terminal establishes the received ultrasonic echo signal. An affine model, using a relaxation algorithm to estimate the parameters of the affine model, and obtain the frequency and amplitude of the blood flow signal.

The color ultrasound parameterized blood flow signal extraction method adopted in this embodiment assumes that the input signal is a mathematical model with unknown parameters, and these unknown parameters (such as frequency, intensity) are obtained from the actually received input signal through a certain algorithm. The Doppler frequency of the final blood flow signal is obtained by appropriate selection of these estimated parameters. The advantage of the parameterized method is that it can not only effectively filter out clutter, but also eliminate the interference of white noise, thereby reducing the variance of frequency estimation. The greater the variance, the worse the effect. Therefore, the color ultrasound parameterized blood flow signal extraction method has advantages in theory and practice for estimating blood flow velocity.

The color ultrasound parameterized blood flow signal extraction method is mainly based on the following two considerations. First, the modeling of the signal; second, the parameter estimation of the model. For signal modeling, one way is the classical pole-zero model. For example, the autoregressive model (Autoregressive, AR) is a method of calculating the Doppler frequency of each principal component by analyzing all poles of the signal. In the AR method, the ultrasonic echo data is expressed as a linear combination of past samples of this signal and white noise. The weight parameters in the model are solved by Burg method or Yule-walker method. The AR model does not consider the high-intensity characteristics of clutter, so when the clutter energy is very strong, the estimation of blood flow signal has a large error. Another widely used modeling method is the feature component model, such as the multiple signal classification method (Multiple signal classification, MUSIC). The method of MUSIC starts from the eigendecomposition of the input signal, and calculates the model parameters through spectral analysis or root finding. In MUSIC, the covariance of the signal needs to be calculated. When the number of samples of the input signal is small, the calculation accuracy is not high.

The invention establishes an affine model for the input signal, and uses a relaxation algorithm to estimate model parameters, aiming at improving the estimated mean value and estimated variance of the Doppler frequency of the blood flow signal.

Embodiment 2. The difference between this embodiment and the color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in Embodiment 1 is that the receiving end establishes an affine model for the received ultrasonic echo signal. The process is:

Use the received ultrasonic echo signal x as an affine model:

x≈Pα (1)

in,

P is the feature matrix:

P=[p(f ₁ ) p(f ₂ ) …p(f _k )…p(f _K )] (2),

The expression of the ultrasonic echo signal x:

x=[x(1) x(2) ... x(n) ... x(N)] ^T (3)

Among them, k is the number of main components, k=1, 2, ..., K, k and K are all positive integers; x(1) represents the first sampling value, x(2) represents the second sampling value , x(n) represents the nth sampling value, x(N) represents the Nth sampling value, n=1,2...,N, N represents the signal length, n and N are natural numbers; p(f _k ) A vector of frequencies representing the kth principal component.

The expression of p(f _k ) is:

$\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; 1</span>}& {\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;</span>}\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}& <span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span>& {\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}\end{array}\right]}^{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

The expression of the amplitude strength vector α is:

α=[α ₁ α ₂ ... α _k ... α _K ] (5)

Based on the established affine model, the calculation problem of Doppler frequency of blood flow signal is transformed into a parameter estimation problem. that is Under the Euclidean norm of , the estimation results of K parameter group parameters {f _k ,α _k ,k=1,…,K} can be solved by the following optimization model, such as

${<span\; class="notranslate">\; \{</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \}</span>}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}{\mathrm{<span\; class="notranslate">\; min</span>}}_{{<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \}</span>}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; P\&alpha;</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Among them, ||·|| represents Euclidean norm of .

When solving the optimization problem described in Equation (5), the frequencies of the principal components are not considered first Instead, focus on the amplitude vector α. That is, assuming that P is known, the least squares solution of the amplitude vector α is obtained,

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; )</span>}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$

Then, use the obtained amplitude vector α to calculate the frequency vector f=[f ₁ f ₂ …f _K ] ^T through the following objective function, such as

${<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \}</span>}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}{\mathrm{<span\; class="notranslate">\; min</span>}}_{{<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \}</span>}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; K</span>}}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; P</span>}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; )</span>}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; P</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

where I is the identity matrix.

Specific Embodiment 3. This embodiment will be described in detail with reference to FIG. 1. The difference between this embodiment and the color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in Embodiment 1 is that the relaxation algorithm is used to estimate the parameters of the affine model. , the specific process of obtaining the frequency and amplitude of the blood flow signal is:

Step A, estimate the frequency and amplitude of the first principal component, subtract the principal component from the input signal, estimate the frequency and amplitude of the second principal component, and perform step B;

Step B. Reversely update the frequency and amplitude of the first principal component according to the frequency and amplitude of the second principal component;

Step C, judging whether the frequencies of the two principal components are stable, if so, execute step D, if not, execute step A;

Step D. Estimate the frequency and amplitude of the third principal component based on the stabilized two principal components, and update the frequency and amplitude of the third principal component based on the frequency and amplitude of the first principal component and the frequency and amplitude of the third principal component. The frequency and amplitude of the two principal components, or update the frequency and amplitude of the first principal component based on the frequency and amplitude of the second principal component and the frequency and amplitude of the third principal component;

Step E, combine the first principal component, the second principal component and the third principal component to obtain the energy of the combined signal, and judge whether the difference between the energy of the combined signal and the energy of the input signal is greater than or equal to greater than 1- δ, δ=10 ^-2 , if yes, select the Doppler frequency with the largest three principal component frequencies as the Doppler frequency of the blood flow signal, if not, perform step A.

Embodiment 4. The difference between this embodiment and the color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in Embodiment 3 is that in step A, the frequency and amplitude of the first principal component are estimated, and the After this principal component is subtracted from the input signal, the specific process of estimating the frequency and amplitude of the second principal component is:

Step 1. Initialize the ultrasonic echo signal x, the ultrasonic echo signal contains K principal components, set K=1, substitute the ultrasonic echo signal x into formulas (9) and (10) to obtain the frequency of the first characteristic component estimated value of and an estimate of the amplitude of the first eigencomponent Execute step two;

$\begin{array}{cc}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ]</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span>$

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>$

Among them, f is a variable, the value range of this variable is [-fs/2, fs/2], fs is the sampling frequency, and the superscript H indicates the conjugate transpose of the vector; is the abscissa value corresponding to the highest peak among all peaks of the periodic spectral function |p(f) ^H x| ² /N; and It can be calculated from the highest peak among all the peaks of the periodic spectral function p(f) ^H x/N.

Step 2. Let K ₀ =K ₀ +1, according to the estimated value of the frequency of the first feature component obtained in Step 1 and an estimate of the amplitude of the first eigencomponent Obtain x ₁ according to formula (11),

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>$

Use formulas (12) and (13) to obtain the estimated value of the frequency of the second characteristic component and an estimate of the amplitude of the 2nd eigencomponent According to estimates and Calculate x ₂ using formula (14),

$\begin{array}{cc}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& \mathrm{<span\; class="notranslate">\; mi</span>}{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ]</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}& {\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 12</span>}<span\; class="notranslate">\; )</span>$

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; N</span>}}{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; =</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 13</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{\mathrm{<span\; class="notranslate">\; K</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&alpha;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; .</span>$

It can be seen in this embodiment

Embodiment 5. The difference between this embodiment and the color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in Embodiment 3 is that in step B, the frequency and amplitude of the second principal component are reversed. The specific process of updating the frequency and amplitude of the first principal component is: According to the obtained _x2 , it is obtained by formulas (9) and (10) and

Embodiment 6. The difference between this embodiment and the color ultrasound parameterized blood flow signal extraction method based on the relaxation algorithm described in Embodiment 3 lies in the specific process of determining whether the frequencies of the two principal components are stable in step C. is: the estimated value with estimates Make a difference, and judge whether the absolute value of the obtained difference is less than or equal to 10 ^-3 , if less than or equal to, the frequency of the two principal components is stable, and if greater, the frequency of the two principal components is unstable.

Specific calculation example:

In the simulated signal synthesized by Doppler mask, the clutter component has relatively low frequency characteristics, but the energy of the signal is far stronger than the blood flow signal and thermal noise. The blood flow signal has a relatively high frequency characteristic. Although the energy of the signal is obviously lower than the clutter component, it is still stronger than the thermal noise. Noise is a low-energy component of a full frequency band, which can be represented by Gaussian white noise in simulation.

In order to verify the wide applicability of the proposed method, some parameters are fixed in the simulation, while the parameters of particular interest are varied. Table 1 lists the parameters used in the simulation and their values. Table 1 shows the experimental parameters.

Table 1

In the simulation and its discussion, the basic unit of all frequency-related values is marked as the pulse repetition frequency (Pulse repetition frequency, PRF). In the simulation, the center frequency of the clutter is 0.1PRF. The center frequency of blood flow varies from 0.05PRF to 0.5PRF with an interval of 0.025PRF. In order to meet the real-time requirements of ultrasound imaging, all tests are performed on the clutter filtering effect under 16 pulse periods. In addition, in order to eliminate systematic errors, 1024 simulations are performed for each set of simulation parameters.

The present invention is compared with two popular parameterized blood flow signal estimation methods: second-order autoregressive model and second-order multiple signal classification based on root-finding method, and the results are shown in Fig. 2 to Fig. 5 . It can be found that the present invention exhibits accurate estimation results in all cases where the Doppler of the blood flow signal is lower than 0.45PRF.

This is because the relaxation algorithm first estimates the signal frequency from the input signal, which is approximately the center frequency of the clutter due to the large clutter intensity; then, the Doppler frequency of the blood flow signal is estimated from the residual signal, As described in step A of the described algorithm. The estimated blood flow signal parameters are in turn applied to update the parameters of the clutter components, as shown in step B. This process is repeated until the Doppler frequencies of the blood flow signal and clutter remain stable, ie the difference between two successive estimates of the step C requirement is below 10 ⁻³ PRF. At this time, calculate the energy ratio of the sum signal of the estimated clutter and blood flow signal to the input signal. Due to the existence of white noise, at this time, turn to estimate the parameters of the third principal component, that is, return to step 2. Repeat the previous steps until the frequencies of the three principal components are stable, as shown in step D. Finally, as described in step E, the Doppler frequency corresponding to the principal component with the second largest intensity parameter is selected as the Doppler frequency output of the blood flow signal.

From the above, accurate estimation of the parameters of the clutter will lead to less clutter components in the residual signal, thereby improving the parameter estimation accuracy of the blood flow signal. Conversely, accurately estimating the parameters of blood flow signals will improve the estimation accuracy of clutter parameters. These two advantages are iteratively added until the estimated parameters stabilize. The final result shows that the dual parameter estimation method of the present invention simultaneously realizes the accurate estimation of the parameters of the clutter and the blood flow signal, especially when the Doppler frequency of the blood flow signal is relatively close to the center frequency of the clutter component , the estimation effect of the present invention is obviously better than the other two parameterized blood flow signal extraction methods.

Claims (5)

1., based on the colorful ultrasonic parametrization blood flow signal extracting method of relaxed algorithm, adopt colorful ultrasonic detector to gather blood flow signal, and be emitted to Received signal strength end; Received signal strength end sets up affine model to the ultrasound echo signal received, and adopts relaxed algorithm to estimate affine model parameter, obtains blood flow signal frequency and amplitude;

It is characterized in that: Received signal strength end to the detailed process that the ultrasound echo signal received sets up affine model is:

Using the ultrasound echo signal x of reception as affine model:

x≈Pα (1)

Wherein, α is oscillator intensity vector,

P is eigenmatrix:

P＝[p(f ₁)p(f ₂)…p(f _k)…p(f _K)] (2)，

The expression formula of ultrasound echo signal x:

x＝[x(1)x(2)…x(n)…x(N)] ^T (3)

Wherein, k is the number of main constituent, k=1,2 ..., K, k and K are positive integer; X (1) represents first sampled value, and x (2) represents second sampled value, and x (n) represents the n-th sampled value, and x (N) represents N number of sampled value, n=1,2 ..., N, n and N are natural number; p (f _k) represent the frequency vector of a kth main constituent.

2. the colorful ultrasonic parametrization blood flow signal extracting method based on relaxed algorithm according to claim 1, is characterized in that: adopt relaxed algorithm to estimate affine model parameter, the detailed process obtaining blood flow signal frequency and amplitude is:

Steps A, the frequency estimating first main constituent and amplitude, after described first main constituent being deducted from input signal, estimate frequency and the amplitude of second main constituent, perform step B;

Step B, upgrade frequency and the amplitude of first main constituent conversely according to the frequency of second main constituent and amplitude;

Step C, judge whether the frequency of two main constituents is stablized, if perform step D, perform steps A if not;

Step D, stablize according to the frequency of two main constituents after the frequency of two Principal Component Estimation the 3rd main constituent and amplitude, upgrade frequency and the amplitude of second main constituent according to the frequency of first main constituent and the frequency of amplitude and the 3rd main constituent and amplitude, or upgrade frequency and the amplitude of first main constituent according to the frequency of second main constituent and the frequency of amplitude and the 3rd main constituent and amplitude;

Step e, signal is carried out in first main constituent, second main constituent and the 3rd main constituent merge the energy obtaining combined signal, judge whether the difference of the energy of this combined signal and the energy of input signal is more than or equal to and be greater than 1-δ, δ=10 ^-2, if so, then select the Doppler frequency as blood flow signal that three main constituent frequencies are maximum, perform steps A if not.

3. the colorful ultrasonic parametrization blood flow signal extracting method based on relaxed algorithm according to claim 2, it is characterized in that: the frequency of estimation first main constituent described in steps A and amplitude, after described first main constituent being deducted from input signal, estimate that the frequency of second main constituent and the detailed process of amplitude are:

Step one, initialization ultrasound echo signal x, this ultrasound echo signal comprises K main constituent, if K=1, by ultrasound echo signal x in formula (9) and (10), obtains the estimated value of the 1st characteristic component frequency with the estimated value of the amplitude of the 1st characteristic component perform step 2;

$\begin{array}{c}{\hat{f}}_1=\arg \min \left|\right|_f[I-\frac{p\left(f\right)p{\left(f\right)}^H}{N}]x{\left|\right|}^2\\ =\arg \min |p{\left(f\right)}^{H}x{|}^2_f\end{array}---\left(9\right)$

${\widehat{\mathrm{\&alpha;}}}_1=\frac{p{\left(f\right)}^{H}x}{N}{|}_{f={\hat{f}}_1}---\left(10\right)$

Wherein, f is variable, and the span of this variable is [-fs/2, fs/2], and fs is sample frequency, and superscript H represents the conjugate transpose of vector;

Step 2, make K ₀=K ₀+ 1, according to the estimated value of the 1st the characteristic component frequency that step one obtains with the estimated value of the amplitude of the 1st characteristic component x is obtained according to formula (11) ₁,

$x_1=x-\underset{k=2}{\overset{K}{\mathrm{\&Sigma;}}}{\widehat{\mathrm{\&alpha;}}}_{k}p\left({\hat{f}}_k\right)---\left(11\right),$

Formula (12) and (13) is adopted to obtain the estimated value of the 2nd characteristic component frequency with the estimated value of the amplitude of the 2nd characteristic component according to estimated value with formula (14) is adopted to calculate x ₂,

$\begin{array}{c}{\hat{f}}_2=\arg \min \left|\right|_f[I-\frac{p\left(f\right)p{\left(f\right)}^H}{N}]x_1{\left|\right|}^2\\ =\arg \min |p{\left(f\right)}^H{x}_1{|}^2_f\end{array}---\left(12\right)$

${\widehat{\mathrm{\&alpha;}}}_2=\frac{p{\left(f\right)}^H{x}_1}{N}{|}_{f={\hat{f}}_2}---\left(13\right)$

$x_2=x-\underset{k=1,k\&NotEqual;2}{\overset{K}{\mathrm{\&Sigma;}}}{\widehat{\mathrm{\&alpha;}}}_{k}p\left({\hat{f}}_k\right)---\left(14\right).$

4. the colorful ultrasonic parametrization blood flow signal extracting method based on relaxed algorithm according to claim 3, is characterized in that: the detailed process of frequency and amplitude that the frequency according to second main constituent described in step B and amplitude upgrade first main constituent is conversely: according to the x obtained ₂obtained by formula (9) and (10) with

5. the colorful ultrasonic parametrization blood flow signal extracting method based on relaxed algorithm according to claim 3, is characterized in that: the detailed process whether frequency judging two main constituents described in step C is stable is: by estimated value with estimated value do difference, judge whether the absolute value of the difference obtained is less than or equal to 10 ^-3if the frequency being less than or equal to expression two main constituents is stablized, if the frequency being greater than expression two main constituents is unstable.