A vascular flow velocity estimation method based on sparse pulses

Abstract

The present invention discloses a method for estimating blood flow velocity by transmitting sparse pulses, and specifically designs a blood flow velocity estimation method based on sparse pulses. The present invention transmits sparse pulses to obtain echo signals, and then performs autocorrelation operation and vectorization on the echo signals to obtain virtual echo signals with equivalent uniform emission and a larger number of pulses. The result of spectral estimation of the echo signals is compared with the result of spectral estimation of the virtual echo signals, and finally the value of blood flow velocity estimation is obtained. The method used in the present invention can obtain high-resolution, low-artifact blood flow velocity estimation results with a small number of pulses, which is beneficial for medical ultrasound systems to further improve the refresh rate and real-time performance of B-mode imaging.

Description

Technical Field

[0001] The invention relates to the technical field of ultrasonic measurement, and in particular to a blood vessel flow velocity estimation method based on sparse pulses. Background Art

[0002] Ultrasonic Doppler is a commonly used non-invasive method for estimating blood flow velocity. This method emits pulsed ultrasonic waves. When the scattered signal encounters a moving object (such as red blood cells), a Doppler frequency deviation will be generated. The blood flow velocity can be measured by receiving and processing the scattered signal. The echo signal is a superposition of the scattering of many moving objects within a range, so it can be regarded as a distributed source. An echo signal at each depth can be regarded as a snapshot sample. Since the blood flow velocity is a function that changes with the depth of the blood vessel cross-section, it is impossible to collect a large number of snapshot signals for processing. It is a short snapshot model. In addition, in the medical ultrasound system, Doppler pulses and B-mode pulses need to be emitted simultaneously. In the system, the B-mode requires a sufficient number of pulses to obtain accurate imaging and a high refresh rate to facilitate real-time diagnosis by doctors, and at the same time, sufficient Doppler pulses are required to make the blood flow velocity measurement have sufficient resolution. Therefore, we need to arrange the emission arrangement of the two reasonably.

[0003] vbvbvbvbvbvbvbvb… Method 1

[0004] vvvvbbbbvvvvvbbbb… Method 2

[0005] There are two methods now. The first is to transmit B-mode pulses (indicated by b) and Doppler pulses (indicated by v) alternately, and the maximum estimable value of blood flow velocity can be expressed by the following formula.

[0006]

[0007] Where c is the speed of the sound wave, f prf is the pulse emission frequency, f0 is the pulse center frequency, and θ is the angle between the blood flow velocity vector and the probe direction, so this method will cause the maximum estimable velocity to be halved. The second method is to let the two emissions be emitted alternately in a block manner, but this will cause velocity holes in certain time intervals, which is not conducive to real-time observation, and the spectrum estimation will produce artifacts due to spectrum diffusion. Existing blood flow velocity estimation can be roughly divided into two categories: 1) using autocorrelation technology or Welch window (i.e. direct fast Fourier transform) to perform spectral estimation to obtain blood flow velocity, but this method has a low resolution due to spectral peak diffusion and other reasons; 2) using compressed sensing and sparse Bayesian methods, due to their high time complexity, it is difficult to meet the real-time requirements of medical ultrasound imaging. Summary of the invention

[0008] The purpose of the present invention is to solve the above-mentioned defects in the prior art and to provide a vascular flow velocity estimation method based on sparse pulses. The method utilizes sparsely emitted echoes to construct virtual uniform linear echoes, uses virtual echoes as data sources for estimation, and obtains high-resolution, low-artifact blood flow velocity estimation results with a small number of pulses.

[0009] The purpose of the present invention can be achieved by adopting the following technical solutions:

[0010] A method for estimating blood vessel flow velocity based on sparse pulses, the method comprising the following steps:

[0011] S1, use the ultrasound probe to transmit sparse Doppler pulses;

[0012] S2. Receive the echo of the ultrasonic wave with an ultrasonic probe, perform sampling to obtain the echo digital signal with the blood flow Doppler information, perform Hilbert transform on the echo digital signal, superimpose the echo digital signal to synthesize into a complex signal, and sample the complex signal synthesized by different sparse Doppler pulse echoes at the same depth to form a group of sparse Doppler pulse signals. The data between different pulses in this group of signals is called slow time sampling, and the sampling at different depths in the same pulse is called fast time sampling.

[0013] S3, performing autocorrelation operation on a set of sparse Doppler pulse signals to obtain an autocorrelation matrix, performing eigenvalue decomposition on the autocorrelation matrix to obtain a signal subspace and a noise subspace, and performing spectrum search to obtain a spectrum peak f1 and a spectrum estimate P1;

[0014] S4, vectorizing the autocorrelation matrix to obtain a vector, sorting the values ​​in the vector, and performing a weighted average operation to form a virtual uniform linear Doppler echo pulse;

[0015] S5, constructing a virtual uniform linear Doppler echo pulse into a Teplitz matrix, performing eigenvalue decomposition on the Teplitz matrix to obtain a virtual uniform pulse signal subspace and a virtual uniform pulse noise subspace, and performing spectrum search using the orthogonality of the flow pattern array and the noise space to obtain a spectrum peak f2 and a spectrum estimate P2;

[0016] S6, compare the difference between the spectrum peak f1 and the spectrum peak f2. When |f1-f2|≤δ, use the spectrum estimate P2 as the spectrum search result. When |f1-f2|>δ, use the spectrum estimate P1 as the spectrum search result. Finally, the spectrum estimate P s (f) Blood flow velocity estimation P is obtained based on the relationship between frequency and blood flow velocity s (v).

[0017] In the case of sparse transmission, the spectrum estimate P1 obtained by eigenvalue decomposition of the echo signal will produce artifacts due to spectral diffusion due to sparseness, and the resolution is low due to insufficient number of pulses. The use of virtual uniform linear Doppler echo pulses can solve the artifacts caused by spectral diffusion due to their uniformity, and more pulses can be virtualized, so the spectrum estimate P2 can obtain a higher resolution. However, due to the limited number of snapshots that can be selected, the error of the constructed virtual uniform linear Doppler echo pulse increases, which may cause the spectrum estimate P2 to have estimation errors. Therefore, by comparing the difference between the spectrum peak f1 and the spectrum peak f2, the spectrum estimate P1 is used to compensate for the error in the spectrum estimate P2, thereby improving the accuracy.

[0018] Furthermore, the first half of the sparse Doppler pulse is composed of a uniform pulse signal, which is used to increase the proportion of the minimum transmission interval and reduce the generation of artifacts. The second half can use a sparse transmission form so that the entire sparse Doppler pulse can construct a virtual uniform linear Doppler echo pulse, that is, the difference set L of the sparse transmission Doppler pulse sequence number diff ={p n -p m |n,m∈[0,P-1]} is a range of -p P-1 to p P-1 The set of elements of the unit uniform linear array, where p n and p m Represent the sequence numbers of the nth and mth transmitted pulses respectively, and p0=0, the sparsity of the sparse Doppler pulse can reach 84% to 88%, and the transmission arrangement is a nested arrangement and a minimum redundant arrangement with a uniform linear array, such as Figure 2 shown.

[0019] Sparsity: used to indicate the sparseness of the pulse, its value is equal to The pulse aperture value is the value of the last pulse number minus the first pulse number.

[0020] Furthermore, in step S2, the echo digital signal is a superposition of the Doppler frequencies of a section of blood flow, and without considering noise, the echo digital signal a[k, p] can be expressed by the following formula.

[0021]

[0022] Where N is the number of scattering particles, α n represents the scattering intensity of the nth scattering particle, f s Represents the sampling frequency. When the data of the echo digital signal is obtained at the kth sampling point, it is equivalent to sampling at a certain depth. The same pulse sampling points are combined to form a vector. At this time, k is also called a snapshot. Since the blood flow velocity is a function that changes with the depth of the blood vessel section, such as Figure 3 As shown in the figure, the flow velocity is fast in the center of the blood vessel and relatively slow near the blood vessel wall, so it is impossible to use snapshots at distant intervals, that is, the range of k is limited, v(k) n represents the velocity of blood flow perpendicular to the pulse direction, f0 is the center frequency of the pulse signal, c is the propagation speed of sound, T prf is the pulse emission period. p is the timing of pulse emission, d0 is the distance from the probe to the center of the blood vessel, and the pulse includes M periods in total.

[0023] The echo digital signal obtained from each pulse is Hilbert transformed and superimposed with the value of the original echo digital signal to obtain a set of sparse Doppler pulse signals, which can be expressed as x = [x[0], x[p1], ..., x[p2], ..., x[p P-1 ]] T , where x[p] is expressed as follows:

[0024]

[0025] where v z =∫vg(v)dv represents the blood flow velocity of the distributed source type, v represents the central velocity of the blood flow of the distributed source type, g(v) represents the diffusion velocity distribution, and e(p) represents the noise.

[0026] Furthermore, in step S3, a set of sparse Doppler pulse signals is used to perform spectrum peak estimation and spectrum search. First, a set of sparse Doppler pulse signals is autocorrelated to obtain an autocorrelation matrix R=E[xx H ], perform eigenvalue decomposition on the autocorrelation matrix, the formula is as follows:

[0027]

[0028] where ∑ S and∑ N They are the signal subspace and the noise subspace, U s and U N are the signal eigenvector and the noise eigenvector respectively. S represents the first C largest eigenvalues, Σ N represents the last PC eigenvalues, P and C are positive integers greater than zero, P can take a value between 20 and 128. For the case of short snapshot distributed sources, the value of C is divided and valued by directly observing the size of the eigenvalue;

[0029] Since the flow pattern array and the noise space are orthogonal, that is, a(f) H U N = 0, where a(f) is a set of flow pattern arrays of sparse Doppler pulse signals, expressed as

[0030] a(f)=[1,exp(πjfp1T prf ),…,exp(2πjp P-1 f prf )]

[0031] According to the spectrum search, the spectrum peak f1 and spectrum estimation P1 are obtained. The spectrum search formula is as follows

[0032]

[0033] Furthermore, the process of step S4 is as follows:

[0034] The autocorrelation matrix is ​​vectorized to obtain the vector z=vec(R), and the vectorized elements correspond to the cross-correlation of a set of slow-time signals, which is expressed as follows:

[0035]

[0036] where p n and p m Represents the serial number of the transmitted pulse, z(p n ,p m ) represents the value after the cross-correlation operation between the n-1th complex signal and the m-1th complex signal, σ is the noise power. I represents the unit matrix, z(p n ,p m ) and the received signal x[p] are in the same form, z(p n ,p m ) is regarded as (p n -p m ) complex signal z(p n -p m ). n -p m ) are rearranged in order, the data at the same position are accumulated and weighted averaged to construct a virtual uniform linear Doppler echo pulse z.

[0037] The virtual uniform linear Doppler echo pulse is used to construct a Teplitz matrix and perform spectrum search. The Teplitz matrix R tep The expression is as follows:

[0038]

[0039] in p n -p m =p,p n and p m Represents the serial number of the transmitted pulse. The sparse Doppler pulses transmitted ensure that a virtual uniform linear Doppler echo pulse can be constructed, that is, p n -p m The value is [-pP-1 ,p P-1 ], for all integers between tep Perform eigenvalue decomposition and we get:

[0040]

[0041] in and are the virtual uniform pulse signal subspace and the virtual uniform pulse noise subspace, and are the virtual uniform pulse signal feature vector and the virtual uniform pulse noise feature vector respectively. represents the first C largest eigenvalues, ∑ N Represents the latter P-1 -C eigenvalues, since the flow pattern array and the virtual uniform impulse noise space are orthogonal, that is, in is the flow pattern array of virtual uniform linear Doppler echo pulses, and the expression is

[0042]

[0043] According to the spectrum search, the spectrum peak f2 and spectrum estimation P2 are obtained. The spectrum search formula is as follows

[0044]

[0045] Furthermore, in the process of step S6, since the value range of the complex signal snapshot k is limited, the construction of the virtual uniform linear Doppler echo pulse z has a larger error than the directly received set of sparse Doppler pulse signals, which causes the spectrum estimation P2 to have an estimation deviation at the point with large noise. Therefore, the difference between the spectrum peak f1 and the spectrum peak f2 is compared. When the spectrum estimation P2 has a deviation, the value of the spectrum estimation P1 is used to compensate for it, and finally the spectrum estimation P is obtained. s (f). The reference value δ of the difference between the peak f1 and the peak f2 is selected in the range of

[0046] The relationship between frequency and speed is as follows:

[0047]

[0048] where v z is the blood flow velocity, c is the speed of sound, and f0 is the center frequency of the pulse signal. s (f) Obtain blood flow velocity estimate P s (v).

[0049] Compared with the prior art, the present invention has the following advantages and effects:

[0050] 1. The present invention uses sparsely transmitted pulses to achieve a higher sparsity, and only a small number of Doppler pulses are needed to obtain blood flow velocity estimation results, allowing the medical ultrasound system to transmit more B-mode pulses to improve the refresh rate of B-mode imaging.

[0051] 2. When the present invention estimates blood flow velocity, it uses virtual uniform linear Doppler echo pulses to construct the Taeplitz matrix, which has significantly reduced complexity compared to using autocorrelation operations. Then, spectral estimation is performed after eigenvalue decomposition to obtain the blood flow velocity estimation result. The use of virtual uniform linear Doppler echo pulses greatly increases the number of pulses, thereby effectively improving the resolution of the estimation. Relying on its uniform linear characteristics, the spectral diffusion is reduced, the artifacts are effectively suppressed, and the effect of blood flow velocity estimation is improved.

[0052] 3. The present invention uses a quadratic search algorithm. During the construction of a virtual uniform linear Doppler echo pulse, due to the limitation of the number of snapshots, the autocorrelation matrix R is only an approximate matrix of the real autocorrelation matrix, thereby increasing the error of the virtual uniform linear Doppler echo pulse, causing it to undergo subsequent Taeplitz matrix and eigenvalue decomposition, and ultimately resulting in a large error when the spectrum estimation value is obtained. Therefore, when a large error occurs, the quadratic search algorithm is used. When the autocorrelation matrix R is used to perform eigenvalue calculations to obtain a spectrum estimation, the error is compensated, thereby improving the accuracy of the estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0054] Figure 1 It is a flow chart of a blood flow velocity estimation calculation method based on sparse pulses disclosed in the present invention;

[0055] Figure 2 is an example diagram of a nested transmission of 30 pulses and a minimum redundant transmission arrangement with a uniform linear array in an embodiment of the present invention;

[0056] Figure 3 is an example diagram of a blood flow model and sampling points in an embodiment of the present invention;

[0057] Figure 4 This is a search result diagram of the minimum redundant transmission arrangement echo signal calculation spectrum of 34 pulses with a uniform linear array in an embodiment of the present invention;

[0058] Figure 5 This is a search result diagram of a calculated spectrum of a virtual echo signal of a minimum redundant transmission arrangement of 34 pulses with a uniform linear array in an embodiment of the present invention;

[0059] Figure 6 This is a diagram showing selected results of a search for a calculated spectrum of a virtual echo signal of a minimum redundant transmission arrangement of 34 pulses with a uniform linear array in an embodiment of the present invention;

[0060] Figure 7 This is a diagram of selected results of a secondary search spectrum search for a minimum redundant transmission arrangement of 34 pulses with a uniform linear array in an embodiment of the present invention.

[0061] Figure 8 This is a graph showing the search results of the calculation spectrum of the echo signal arrangement of 30 pulse nested transmissions in an embodiment of the present invention;

[0062] Fig. 9 This is a graph of the search results of the secondary search spectrum of 30 pulse nested emission arrangements in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] Example 1

[0065] The flow chart of the blood flow velocity estimation calculation method in this embodiment is as follows: Figure 1 This embodiment uses matlab as the simulation platform and Field II for simulation to establish a blood flow motion model. Figure 3 As shown. Set the transmission signal, calculate the echo digital signal, and estimate the blood flow velocity. According to the time-frequency (velocity) diagram of the blood flow velocity estimation, the performance of the method in the present invention is verified. The specific implementation steps are as follows:

[0066] T1, transmit sparse pulse signal

[0067] The minimum redundancy arrangement with a uniform linear array of 34 pulses is as follows: Figure 2 As shown in FIG. 1 , the number of initial partial uniform pulse signals is 16, and the sparsity is 84%. The minimum redundant arrangement can construct the longest virtual uniform linear Doppler echo pulse with the least number of sparse pulses, and the transmission duration is 1 second.

[0068] T2, echo signal processing

[0069] In this embodiment, a probe is used to transmit a sinusoidal pulse wave, and the pulse is collected after being reflected by particles such as red blood cells in the blood vessels. The collected echo digital signal a[k, p] can be expressed by the following formula:

[0070]

[0071] Where M = 4, f s =100MHz, c=1540m / s, f0=1MHz, d0=6cm, T prf =5KHz. The phase change between adjacent snapshots is 2πf0 / f s =0.02π, so the data between adjacent snapshots are basically the same, and a total of 31 snapshots are selected. The echo digital signal a[k,p] is Hilbert transformed, and the echo signals are superimposed to synthesize a complex signal. The complex signals obtained by different sparse Doppler pulse echoes are sampled at the same depth to form a set of sparse Doppler pulse signals, which can be expressed as x=[x[0],x[p1],…,x[p2],…,x[p P-1 ]] T , there are 34 sub-elements in total, where the expression of x[p] is as follows:

[0072]

[0073] Autocorrelation of a set of sparse Doppler pulse signals is performed to obtain an autocorrelation matrix R with a dimension of 34×34.

[0074] T3. Echo signal calculation

[0075] Perform eigenvalue decomposition on the autocorrelation matrix R, sort the eigenvalues, and observe the eigenvalues ​​to find that there is a largest eigenvalue, which means that the signal is concentrated in this eigenvalue. Therefore, take the first eigenvalue as the signal subspace ∑ S , signal subspace ∑ S The corresponding eigenvector is U s , take the last 33 eigenvalues ​​as the noise subspace ∑ N , noise subspace ∑ N The corresponding eigenvector is U N .

[0076] According to the orthogonal relationship between the flow pattern array and the noise space, the spectrum search is performed as follows:

[0077]

[0078] Where a(f) is a set of sparse Doppler pulse signal flow arrays with a dimension of 1×34, expressed as:

[0079] a(f)=[1,exp(2πjfT prf ),exp(4πjfT prf )…,exp(516πjfT prf )]

[0080] The maximum value is obtained by searching and is counted as the spectrum peak f1. The spectrum estimation P1 is as follows: Figure 4 As shown, the vertical axis frequency f can be based on It is converted into blood flow velocity, thus becoming the result of blood flow velocity estimation. As can be seen from the figure, using a set of sparse transmitted pulse signals to directly perform spectrum estimation has obvious artifacts and the estimation result is poor.

[0081] T4. Virtual echo signal calculation

[0082] The autocorrelation matrix R is vectorized to obtain a vector z=vec(R) with a dimension of 1×1156. Each element in the vector can be regarded as a virtual transmission pulse, and its corresponding transmission timing is (p n -p m ), n and m represent the nth row and mth column of the element in the original autocorrelation matrix. After z is rearranged in order, the data at the same position are accumulated and averaged to obtain a virtual uniform linear Doppler echo pulse with 417 pulses. Use it to construct the Teplitz matrix R tep , as shown below,

[0083]

[0084] in Assuming that the complex signal with the transmission time sequence is p, R tep Perform eigenvalue decomposition, sort the eigenvalues, and take the maximum eigenvalue as the virtual uniform pulse signal subspace Virtual uniform pulse signal subspace The corresponding eigenvector is The last 207 eigenvalues ​​are taken as the virtual uniform impulse noise subspace Virtual Uniform Impulse Noise Subspace The corresponding eigenvector is

[0085] According to the orthogonal relationship between the flow pattern array and the noise space, the spectrum search is performed as follows:

[0086]

[0087] in is the flow pattern array of virtual uniform linear Doppler echo pulses, with a dimension of 1×209, and the expression is:

[0088]

[0089] The maximum value is obtained by searching and is counted as the spectrum peak f2. The spectrum estimation P2 is as follows: Figure 5 As shown, the vertical axis frequency f can be based on It is converted into blood flow velocity, thus becoming the result of blood flow velocity estimation. As can be seen from the figure, compared with Figure 4 The spectrum estimation result obviously suppresses the artifacts and has a higher resolution. However, due to the limited number of snapshots that can be selected, the spectrum estimation of the constructed virtual uniform linear Doppler echo pulse will produce errors at some moments.

[0090] T5. Secondary Search

[0091] In the case of sparse transmission, a set of sparse transmission pulse signals is used to perform autocorrelation operation and then eigenvalue decomposition to obtain the spectrum estimate P1. Due to the sparseness, the spectrum spreads and thus produces artifacts. Due to the insufficient number of pulses, the resolution is low. The spectrum estimate P2 obtained after autocorrelation operation and eigenvalue decomposition is The uniformity of the More pulses are virtualized to obtain higher resolution, but since the number of snapshots that can be taken by complex signals is limited, this leads to The error will increase. Therefore, the difference between the spectrum peak f1 and the spectrum peak f2 is determined. When |f1-f2|≤δ, P2 is used as the spectrum search result. When |f1-f2|>δ, P1 is used as the spectrum search result. In this embodiment, δ is 293Hz.

[0092] Figure 6 for Figure 5 The excerpt of the search results after 0.8s, Figure 7 This is the excerpt after using the secondary search 0.8s. From the figure we can see that, Figure 6 There are obvious estimation errors in the model, and the quadratic search algorithm successfully compensates for these estimation errors.

[0093] Example 2

[0094] In this embodiment, the blood vessel flow velocity estimation method is implemented by referring to the steps of embodiment 1, wherein in step T1, a nested transmission arrangement of 30 pulses is transmitted as follows: Figure 2 As shown, the number of initial partial uniform pulse signals is 16, the sparsity is 88%, and the emission duration is 1 second.

[0095] In step T2, ... In this embodiment, a probe is used to emit a sinusoidal pulse wave, and the pulse is collected after being reflected by particles such as red blood cells in the blood vessels. The collected echo digital signal a[k, p] can be expressed by the following formula:

[0096]

[0097] Where M = 6, fs =80MHz, c=1540m / s, f0=1MHz, d0=5cm, T prf =5KHz. The phase change between adjacent snapshots is 2πf0 / f s =0.016π, so the data between adjacent snapshots are basically consistent, and a total of 33 snapshots are selected. The complex signals obtained by sampling different sparse Doppler pulse echoes at the same depth form a group of sparse Doppler pulse signals, which can be expressed as x = [x[0], x[p1],…, x[p2],…, x[p P-1 ]] T , there are 30 sub-elements in total, where the expression of x[p] is as follows:

[0098]

[0099] Autocorrelation of a set of sparse Doppler pulse signals is performed to obtain an autocorrelation matrix R with a dimension of 30×30.

[0100] In step T3, the autocorrelation matrix is ​​decomposed by eigenvalues, the eigenvalues ​​are sorted, and the first eigenvalue is taken as the signal subspace ∑ S , take the last 29 eigenvalues ​​as the noise subspace ∑ N , noise subspace ∑ N The corresponding eigenvector is U N According to the orthogonal relationship between the flow pattern array and the noise space, a spectrum search is performed to obtain the spectrum peak f1, and the spectrum estimation P1 is as follows: Figure 8 The vertical axis frequency f can be calculated based on Converted to blood flow velocity, it can be seen that the spectrum estimation P1 has obvious artifacts.

[0101] In step T4, the autocorrelation matrix R is vectorized to obtain a vector z=vec(R) with a dimension of 1×900. After z is rearranged in order, the data at the same position are accumulated and averaged to obtain a virtual uniform linear Doppler echo pulse with 481 pulses. Use it to construct the Teplitz matrix R tep , as shown below,

[0102]

[0103] R tep Perform eigenvalue decomposition, sort the eigenvalues, and take the maximum eigenvalue as the virtual uniform pulse signal subspace The last 207 eigenvalues ​​are taken as the virtual uniform impulse noise subspace Virtual Uniform Impulse Noise Subspace The corresponding eigenvector is According to the orthogonal relationship between the flow pattern array and the noise space, a spectrum search is performed to obtain the spectrum peak f2 and the spectrum estimate P2.

[0104] In step T5, the difference between the peak f1 and the peak f2 is determined. When |f1-f2|≤δ, P2 is used as the spectrum search result. When |f1-f2|>δ, P1 is used as the spectrum search result to obtain the final spectrum estimate P s (f), in this embodiment δ is 280 Hz. Spectral estimate P s (f) Fig. 9 As shown, the vertical axis frequency f can be based on Converted to blood flow velocity, thus becoming the blood flow velocity estimate P s (v) The result shows that the secondary search method has good artifact suppression effect and high resolution advantages.

[0105] In summary, the method of the present invention for transmitting sparse pulses can use a pulse form with high sparsity to achieve very high resolution and have high estimation accuracy, while effectively suppressing artifacts, which has certain advantages over other methods currently proposed.

[0106] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A vascular flow velocity estimation method based on sparse pulses, characterized in that: The blood vessel flow velocity estimation method comprises the following steps: S1, use the ultrasound probe to transmit sparse Doppler pulses; S2. Use an ultrasonic probe to receive ultrasonic echoes, perform sampling to obtain an echo digital signal with blood flow Doppler information, perform Hilbert transform on the echo digital signal, superimpose the echo digital signal to synthesize a complex signal, and sample the complex signal synthesized by different sparse Doppler pulse echoes at the same depth to form a group of sparse Doppler pulse signals. The data between different pulses in this group of signals is called slow time sampling, and the sampling at different depths in the same pulse is called fast time sampling. S3, performing autocorrelation operation on a set of sparse Doppler pulse signals to obtain an autocorrelation matrix, performing eigenvalue decomposition on the autocorrelation matrix to obtain a signal subspace and a noise subspace, and performing spectrum search to obtain a spectrum peak f1 and a spectrum estimate P1; S4, vectorizing the autocorrelation matrix to obtain a vector, sorting the values ​​in the vector, and performing a weighted average operation to form a virtual uniform linear Doppler echo pulse; S5, constructing a virtual uniform linear Doppler echo pulse into a Teplitz matrix, performing eigenvalue decomposition on the Teplitz matrix to obtain a virtual uniform pulse signal subspace and a virtual uniform pulse noise subspace, and performing spectrum search using the orthogonality of the flow pattern array and the noise space to obtain a spectrum peak f2 and a spectrum estimate P2; S6. Compare the difference between the spectrum peak f1 and the spectrum peak f2. When |f1-f2|≤δ, use the spectrum estimate P2 as the spectrum search result. When |f1-f2|>δ, use the spectrum estimate P1 as the spectrum search result. Finally, the spectrum estimate P s (f) The result of blood flow velocity estimation is converted from the result of spectrum estimation to obtain the blood flow velocity estimation P according to the relationship between frequency and blood flow velocity. s (v) The process is as follows: The reference value δ of the difference between the peak f1 and the peak f2 is selected in the range of The relationship between frequency and speed is as follows: where v z is the blood flow velocity, c is the speed of sound, f0 is the center frequency of the pulse signal, and the spectrum estimation P s (f) Transformation to obtain blood flow velocity estimation P s (v); The first half of the sparse Doppler pulse is composed of a uniform pulse signal, and the second half uses a sparse transmission form so that the entire sparse Doppler pulse can construct a virtual uniform linear Doppler echo pulse, that is, the difference set L of the sparse transmission Doppler pulse sequence number. diff ={p n -p m |n,m∈[0,P-1]} is a range of -p P-1 to p P-1 The set of elements of the unit uniform linear array, where p n and p m represent the serial numbers of the nth and mth transmitted pulses respectively, and p0=0, p P-1 represents the serial number of the P-1th transmitted pulse; the arrangement of the sparse Doppler pulse transmission is a nested arrangement or a minimum redundancy arrangement with a uniform linear array; Wherein, in step S2, the echo digital signal is the superposition of the Doppler frequency of a section of blood flow. Without considering the noise, the echo digital signal a[k, p] can be expressed by the following formula: Where N is the number of scattering particles, α n represents the scattering intensity of the nth scattering particle, f s Represents the sampling frequency. When the data of the echo digital signal is obtained at the kth sampling point, it is equivalent to sampling at a certain depth. The same pulse sampling points are combined to form a vector. At this time, k is also called a snapshot, v(k) n represents the velocity of blood flow perpendicular to the pulse direction, f0 is the center frequency of the pulse signal, c is the propagation speed of sound, T prf is the pulse emission period, p is the timing of pulse emission, d0 is the distance from the probe to the center of the blood vessel, and the pulse includes M periods in total; The echo digital signal obtained from each pulse is Hilbert transformed and superimposed with the value of the original echo digital signal to obtain a set of sparse Doppler pulse signals represented as x = [x[0], x[p1], x[p2], ..., x[p P-1 ]] T , where x[p] is expressed as follows: v z =∫vg(v)dv represents the blood flow velocity of the distributed source type, where v represents the central velocity of the blood flow of the distributed source type, g(v) represents the diffusion velocity distribution, and e(p) represents the noise.

2. The method for estimating blood vessel flow velocity based on sparse pulses according to claim 1, characterized in that: In step S3, a set of sparse Doppler pulse signals is used to perform spectrum peak estimation and spectrum search. First, a set of sparse Doppler pulse signals is autocorrelated to obtain an autocorrelation matrix R=E[xx H ], perform eigenvalue decomposition on the autocorrelation matrix, the formula is as follows: where ∑ S and∑ N are the signal subspace and the noise subspace respectively, U s and U N are the signal eigenvector and the noise eigenvector respectively, where Σ S represents the first C largest eigenvalues, Σ N represents the last PC eigenvalues, P and C are positive integers greater than zero, P ≥ 20. For the case of short snapshot distributed sources, the value of C is divided according to the size of the eigenvalue; Since the flow pattern array and the noise space are orthogonal, that is, a(f) H U N = 0, where a(f) is a set of flow pattern arrays of sparse Doppler pulse signals, expressed as a(f)=[1,exp(2πjp1fT prf ),…,exp(2πjp P-1 fT prf )] Among them, p1, ..., p P-1 represents the serial number of the 1st, ..., P-1th transmitted pulse, f represents the frequency of the sparse Doppler pulse signal, and the spectrum peak f1 and spectrum estimate P1 are obtained by spectrum search. The spectrum search formula is as follows 3. The method for estimating blood vessel flow velocity based on sparse pulses according to claim 2, characterized in that: The process of step S4 is as follows: The autocorrelation matrix is ​​vectorized to obtain the vector z=vec(R), and the vectorized elements correspond to the cross-correlation of a set of slow-time signals, which is expressed as follows: where p n and p m represents the serial number of the transmitted pulse, z(p n ,p m ) represents the value after the cross-correlation operation between the n-1th complex signal and the m-1th complex signal, σ is the noise power, I represents the unit matrix, z(p n ,p m ) and the received signal x[p] are in the same form, z(p n ,p m ) is regarded as (p n -p m ) complex signal z(p n -p m ), z(p n -p m ) are rearranged in order, the data at the same position are accumulated and weighted averaged to construct a virtual uniform linear Doppler echo pulse z.

4. The method for estimating blood vessel flow velocity based on sparse pulses according to claim 3, characterized in that: The process of step S5 is as follows: The virtual uniform linear Doppler echo pulse is used to construct a Teplitz matrix and perform spectrum search. The Teplitz matrix R tep The expression is as follows: in p n -p m =p,p n and p m Represents the serial number of the transmitted pulse. The sparse Doppler pulses transmitted ensure that a virtual uniform linear Doppler echo pulse can be constructed, that is, p n -p m The value is [-p P-1 ,p P-1 ], for all integers between tep Perform eigenvalue decomposition and we get: in and are the virtual uniform pulse signal subspace and the virtual uniform pulse noise subspace, and are the virtual uniform pulse signal feature vector and the virtual uniform pulse noise feature vector, represents the first C largest eigenvalues, ∑ N Represents the latter P-1 -C eigenvalues, since the flow pattern array and the virtual uniform impulse noise space are orthogonal, that is, in is the flow pattern array of virtual uniform linear Doppler echo pulses, expressed as According to the spectrum search, the spectrum peak f2 and spectrum estimation P2 are obtained. The spectrum search formula is as follows:

Images