Doppler blood flow noise suppression method and system suitable for digestive tract endoscopic examination

Abstract

The present application relates to a kind of Doppler blood flow noise suppression method and system suitable for digestive tract endoscopic detection.It includes: providing endoscopic detection device, and configure the endoscopic detection device to be detected digestive tract is carried out Doppler ultrasound detection in endoscopic mode, to generate endoscopic scanning echo signal group after Doppler ultrasound detection, the endoscopic echo synthesis signal in endoscopic scanning echo signal group is respectively subjected to adaptive filtering, and generate endoscopic synthesis filtered signal after adaptive filtering, and based on all endoscopic synthesis filtered signal generation endoscopic synthesis filtered signal group, the endoscopic synthesis filtered signal group described above is carried out spectral estimation based on feature representation, to generate Doppler spectrum signal of current Doppler ultrasound detection after spectral estimation.The present application can effectively realize Doppler blood flow noise suppression in digestive tract endoscopy, improve the precision and reliability of Doppler blood flow detection.

Description

Technical Field

[0001] This invention relates to a noise suppression method and system, and more particularly to a Doppler blood flow noise suppression method and system suitable for gastrointestinal endoscopy. Background Technology

[0002] In the field of gastrointestinal endoscopic ultrasound imaging, large-size ring array probes are currently widely used for grayscale imaging and Doppler flow imaging. Microprobe catheters, due to their compact structure and ease of endoscopic insertion, are more suitable for future endoscopic ultrasound diagnosis than ring array probes. However, currently commercially available microprobe catheters are all single-element or dual-element mechanically rotated imaging devices, which cannot achieve Doppler flow detection. Furthermore, endoscopic Doppler flow imaging based on ring array probes is susceptible to motion artifacts and background noise caused by respiratory pull or probe movement.

[0003] In gastrointestinal endoscopic imaging, one-dimensional phased array microcatheters with a center frequency of around 8 MHz can also be used. The diameter of a one-dimensional phased array microcatheter is generally no more than 2.7 mm. The array elements within the phased array are arranged along the length of the catheter, thereby enabling Doppler blood flow imaging. However, similar to ring array probes with a center frequency of around 8 MHz, Doppler blood flow imaging using one-dimensional phased array microcatheters in gastrointestinal imaging is also susceptible to motion artifacts and background noise caused by respiratory pull or probe movement. This results in numerous blood flow artifacts, making it difficult to meet the accuracy requirements of practical Doppler blood flow imaging. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a Doppler blood flow noise suppression method and system suitable for gastrointestinal endoscopy, which can effectively suppress Doppler blood flow noise in gastrointestinal endoscopy and improve the accuracy and reliability of Doppler blood flow detection.

[0005] According to the technical solution provided by the present invention, a Doppler blood flow noise suppression method suitable for gastrointestinal endoscopy is provided, the Doppler blood flow noise suppression method comprising:

[0006] An endoscopic detection device is provided, and the endoscopic detection device is configured to perform Doppler ultrasound detection on the digestive tract to be examined in an endoscopic manner, so as to generate an endoscopic scan echo signal group after Doppler ultrasound detection, wherein the endoscopic scan echo signal group includes several endoscopic echo composite signals.

[0007] Adaptive filtering is applied to the endoscopic echo synthesized signals within the endoscopic scan echo signal group, and an endoscopic synthesized filtered signal is generated after adaptive filtering. Furthermore, an endoscopic synthesized filtered signal group is generated based on all the endoscopic synthesized filtered signals.

[0008] When adaptively filtering each endoscopic echo synthesized signal, at least the in-phase component and quadrature component corresponding to each sampling point in the current endoscopic echo synthesized signal must be determined.

[0009] The above-mentioned endoscopic synthesized and filtered signal group is subjected to spectral estimation based on feature representation, so as to generate the Doppler spectrum signal of the current Doppler ultrasound detection.

[0010] When performing adaptive filtering on the synthesized endoscopic echo signal, at least one sampling point component extraction process is performed on each sampling point within the synthesized endoscopic echo signal. This extraction process generates in-phase and quadrature components corresponding to each sampling point.

[0011] When performing sample point component extraction processing, it includes performing component update calculation processing a preset number of times, wherein,

[0012] When performing component update calculations, the following steps are included:

[0013] At least the component array to be updated and the frequency depth information representing the current sampling point depth state are obtained, wherein the frequency depth information includes a first frequency depth component and a second frequency depth component, and the sum of squares of the first frequency depth component and the second frequency depth component is 1.

[0014] The frequency depth information is used to calculate and update the acquired component array, and after the calculation and update, the component array to be updated is generated for the next component update calculation.

[0015] After performing the component update calculation a preset number of times, the corresponding calculated and updated component array is configured as the component extraction target array.

[0016] Based on the component extraction target array, the in-phase component and the orthogonal component corresponding to each sampling point are determined.

[0017] When using frequency depth information to calculate and update the acquired component array, we have:

[0018]

[0019] Where F represents frequency depth information, and P is the component array to be updated. T Let F be the transpose of the frequency depth information, x(n) be the amplitude information of the current sampling point, mu be the convergence factor, and λ be the calculation update coefficient. This refers to the component array obtained after performing component update calculations. The dimension of the component array P to be updated is consistent with the dimension of the frequency depth information F.

[0020] For the frequency depth information of the current sampling point, we have:

[0021]

[0022] Where F represents the frequency depth information, s represents the first component of the frequency depth, c represents the second component of the frequency depth, f_mod represents the demodulation frequency, fs represents the echo sampling frequency, and n represents the sampling depth position of the current sampling point within the corresponding endoscopic echo synthesized signal.

[0023] When performing spectral estimation based on feature representations, the following steps are included:

[0024] Based on the filtered signal group synthesized by endoscopy, the corresponding autocorrelation matrix R is constructed, and the noise subspace matrix G of the autocorrelation matrix R is calculated.

[0025] Construct a steering vector with normalized frequency, and calculate the corresponding Doppler spectrum signal of the current endoscopic synthesized and filtered signal group based on the constructed steering vector and the noise subspace matrix G.

[0026] For the constructed autocorrelation matrix R, we have:

[0027]

[0028] Where N is the number of endoscopic synthesized and filtered signals in the endoscopic synthesized and filtered signal group, L is the number of valid samples, and M is the number of observation samples set, L = N - M + 1, [x1(i)] HT Let x1(i) be the conjugate transpose of x1(i), S(i) be the synthesized and filtered signal of the i-th endoscope, and S(i-1) be the synthesized and filtered signal of the (i-1)-th endoscope.

[0029] When calculating the noise subspace matrix G of the autocorrelation matrix R, we have:

[0030] The autocorrelation matrix R is subjected to eigenvalue decomposition to obtain the eigenvalues ​​of the autocorrelation matrix R and the eigenvector corresponding to each eigenvalue;

[0031] Sort the eigenvalues ​​of the autocorrelation matrix R in descending order, select the eigenvector components corresponding to the (k+1)th eigenvalue to the Mth eigenvalue, and generate the noise subspace matrix G based on the selected eigenvector components.

[0032] When constructing the steering vector of the normalized frequency, we have:

[0033]

[0034] Among them, f g For the normalized frequency, f g The value range is 0 to 1, and the normalized frequency f gThe step length is 1 / M;

[0035] When calculating the generated Doppler spectrum signal, we have:

[0036]

[0037] Where y(f) is the Doppler spectral signal, G T Let (aw) be the transpose of the noise subspace matrix G. T It is the transpose of the guiding vector.

[0038] When performing Doppler ultrasound examination on the digestive tract, it includes several sequential ultrasound scans. Each ultrasound scan includes sequential excitation emission processing, echo acquisition processing, and beamforming processing.

[0039] During the excitation transmission process, each array element in the endoscopic detection device transmits an ultrasonic signal into the digestive tract to be examined.

[0040] During echo acquisition and processing, each array element in the endoscopic detection device is configured to receive the ultrasound echo signal reflected from the digestive tract to be examined;

[0041] When performing beamforming, the ultrasonic echo signals received by all array elements are beamformed to generate a single endoscopic echo composite signal.

[0042] The number of synthesized endoscopic echo signals in the endoscopic scan echo signal group is consistent with the number of times the excitation emission process is performed.

[0043] A Doppler blood flow noise suppression system suitable for gastrointestinal endoscopic examination includes an endoscopic detection device and an endoscopic processor adapted and connected to the endoscopic detection device, wherein...

[0044] When performing Doppler ultrasound examination on the digestive tract, the endoscope processor and the endoscope detection device use the Doppler blood flow noise suppression method described above to suppress noise.

[0045] The advantages of this invention are as follows: Endoscopic echo signal groups generated by Doppler ultrasound detection of the digestive tract are adaptively filtered for each synthesized endoscopic echo signal to determine the in-phase and quadrature components corresponding to each sampling point within the current synthesized endoscopic echo signal; after filtering, the signal group is subjected to spectral estimation based on feature representation, which effectively distinguishes noise from blood flow signals. That is, by utilizing the characteristic differences between noise and blood flow signals, noise and blood flow signals can be accurately separated, thereby effectively suppressing blood flow noise. This solves the problem of blood flow noise in digestive tract endoscopic imaging in this technical field, improving the accuracy and reliability of Doppler blood flow detection. Attached Figure Description

[0046] Figure 1 This is a flowchart of one embodiment of the Doppler blood flow noise suppression of the present invention.

[0047] Figure 2 This is a flowchart of an embodiment of the present invention for adaptive filtering of endoscopic echo synthesis signals.

[0048] Figure 3 This is a structural block diagram of one embodiment of the Doppler blood flow noise suppression of the present invention. Detailed Implementation

[0049] The present invention will be further described below with reference to specific accompanying drawings and embodiments.

[0050] To effectively suppress Doppler blood flow noise during gastrointestinal endoscopy and improve the accuracy and reliability of Doppler blood flow detection, this invention provides a Doppler blood flow noise suppression method suitable for gastrointestinal endoscopy. Specifically, the Doppler blood flow noise suppression method includes:

[0051] An endoscopic detection device is provided, and the endoscopic detection device is configured to perform Doppler ultrasound detection on the digestive tract to be examined in an endoscopic manner, so as to generate an endoscopic scan echo signal group after Doppler ultrasound detection, wherein the endoscopic scan echo signal group includes several endoscopic echo composite signals.

[0052] Adaptive filtering is applied to the endoscopic echo synthesized signals within the endoscopic scan echo signal group, and an endoscopic synthesized filtered signal is generated after adaptive filtering. Furthermore, an endoscopic synthesized filtered signal group is generated based on all the endoscopic synthesized filtered signals.

[0053] When adaptively filtering each endoscopic echo synthesized signal, at least the in-phase component and quadrature component corresponding to each sampling point in the current endoscopic echo synthesized signal must be determined.

[0054] The above-mentioned endoscopic synthesized and filtered signal group is subjected to spectral estimation based on feature representation, so as to generate the Doppler spectrum signal of the current Doppler ultrasound detection.

[0055] It should be understood that when performing Doppler blood flow noise suppression, Doppler ultrasound examination of the digestive tract should be performed. When performing Doppler ultrasound examination of the digestive tract, an endoscopic examination device should be provided, and the endoscopic examination device should be used to perform Doppler ultrasound examination of the digestive tract in an endoscopic manner. Therefore, the provided endoscopic examination device should meet the requirements for Doppler ultrasound examination of the digestive tract. As can be seen from the background art description, the endoscopic examination device can be a ring array probe or a one-dimensional phased array microcatheter, and the endoscopic examination device is preferably a one-dimensional phased array microcatheter.

[0056] Understandably, once an endoscopic examination device is selected, Doppler ultrasound examination of the digestive tract can be performed using existing methods, and the methods and procedures for Doppler ultrasound examination of the digestive tract can be consistent with existing technologies. Figure 1 It is known that after performing Doppler ultrasound examination on the digestive tract, at least an endoscopic scan echo signal group should be generated. The method of generating the endoscopic scan echo signal group can be consistent with existing technologies. The endoscopic scan echo signal group should generally include multiple endoscopic echo composite signals. The following is a detailed explanation of the method of using an endoscopic detection device to perform Doppler ultrasound examination on the digestive tract and generating an endoscopic scan echo signal group.

[0057] In one embodiment of the present invention, when performing Doppler ultrasound detection on the digestive tract, it includes several sequential ultrasound scanning processes, wherein each ultrasound scanning process includes sequential excitation emission processing, echo acquisition processing, and beamforming processing.

[0058] During the excitation transmission process, each array element in the endoscopic detection device transmits an ultrasonic signal into the digestive tract to be examined.

[0059] During echo acquisition and processing, each array element in the endoscopic detection device is configured to receive the ultrasound echo signal reflected from the digestive tract to be examined;

[0060] When performing beamforming, the ultrasonic echo signals received by all array elements are beamformed to generate a single endoscopic echo composite signal.

[0061] The number of synthesized endoscopic echo signals in the endoscopic scan echo signal group is consistent with the number of times the excitation emission process is performed.

[0062] Specifically, the digestive tract to be examined is the digestive tract to be subjected to Doppler ultrasound examination. When performing Doppler ultrasound examination on the digestive tract to be examined, multiple sequential ultrasound scans should be performed using an endoscopic examination device. Ultrasound scans can be performed at a certain frequency, which can be selected as needed to meet the requirements of Doppler ultrasound examination of the digestive tract. In practice, each ultrasound scan is performed identically. The ultrasound scan process may include sequential excitation emission processing, echo acquisition processing, and beamforming processing. The specific methods of excitation emission processing, echo acquisition processing, and beamforming processing are explained below.

[0063] As explained above, the endoscopic detection device contains array elements. The arrangement of these elements depends on the type of endoscopic detection device. For example, if the endoscopic detection device is a one-dimensional phased array microcatheter, the array elements are distributed along the length of the one-dimensional phased array microcatheter. When performing excitation emission processing, specifically, the array elements within the endoscopic detection device are excited to simultaneously emit ultrasonic signals into the digestive tract. Specifically, the method by which the array elements within the endoscopic detection device emit ultrasonic signals into the digestive tract can be consistent with existing methods, aiming to meet the requirements of Doppler ultrasound detection; this will not be elaborated further here.

[0064] After the excitation emission process, echo acquisition processing should be performed to receive the echo signals reflected from the digestive tract. Specifically, during echo acquisition processing, each element within the endoscopic detection device receives the ultrasonic echo signal reflected from the digestive tract being examined. The sampling frequency must satisfy the Nyquist criterion (≥2 times the highest Doppler frequency) when acquiring the echo signal. It should be understood that during echo acquisition processing, ultrasonic echo signals consistent with the number of elements within the endoscopic detection device can be obtained. When the sampling frequency is determined, each ultrasonic echo signal includes the same number of sampling points.

[0065] Each array element in the endoscopic detection device receives an ultrasonic echo signal and can perform beamforming processing. When performing beamforming, the ultrasonic echo signals received by all array elements should be beamformed to obtain a composite endoscopic echo signal. Generally, before performing beamforming processing, the ultrasonic echo signal can also be bandpass filtered. When performing bandpass filtering, a bandpass filter can be designed according to the center frequency of the phased array in the endoscopic detection device. The designed bandpass filter should be designed to effectively suppress high-frequency electronic noise and low-frequency environmental interference.

[0066] In practical implementation, when performing beamforming on all ultrasonic echo signals, one feasible approach is as follows:

[0067]

[0068] Where A(t) is the synthesized endoscopic echo signal, nn is the number of array elements in the endoscopic detection device, and β i Let τ be the i-th weighting coefficient. i Let be the time difference between the arrival of the i-th array element at the synthesis point, t be the synthesis time, and a be the time difference between the arrival of the i-th array element at the synthesis point. i The signal is the ultrasonic echo received by the i-th array element.

[0069] Specifically, the positional distribution of the array elements can be determined based on their arrangement within the endoscopic detection device, thus identifying the i-th array element and the i-th weighting coefficient β. iThe Hanning window coefficient can be used, or other methods can be employed, depending on the specific needs. After determining the synthesis time t, the time difference τ between the i-th array element and the synthesis point can be determined based on the position of the array elements. i .

[0070] As explained above, each sequential ultrasound scan generates one composite endoscopic echo signal. Generally, multiple ultrasound scans should be performed during Doppler ultrasound scanning to obtain multiple composite endoscopic echo signals. The number of composite endoscopic echo signals corresponds to the number of ultrasound scans performed during Doppler ultrasound detection. For example, if 64 ultrasound scans are performed during Doppler ultrasound detection, 64 composite endoscopic echo signals can be obtained.

[0071] Depend on Figure 1 As can be seen, after performing Doppler ultrasound detection and generating an endoscopic scan echo signal group, in order to achieve blood flow noise suppression, each endoscopic echo synthesized signal should be adaptively filtered. Each endoscopic echo synthesized signal can generate an endoscopic synthesized filtered signal after adaptive filtering, and an endoscopic synthesized filtered signal group can be generated based on all the endoscopic synthesized filtered signals. The following is a detailed explanation of the method and process of adaptive filtering for each endoscopic echo synthesized signal.

[0072] In one embodiment of the present invention, when performing adaptive filtering on the endoscopic echo synthesized signal, at least one sampling point component extraction process is performed on each sampling point in the endoscopic echo synthesized signal to generate an in-phase component and a quadrature component corresponding to each sampling point after the sampling point component extraction process, wherein,

[0073] When performing sample point component extraction processing, it includes performing component update calculation processing a preset number of times, wherein,

[0074] When performing component update calculations, the following steps are included:

[0075] At least the component array to be updated and the frequency depth information representing the current sampling point depth state are obtained, wherein the frequency depth information includes a first frequency depth component and a second frequency depth component, and the sum of squares of the first frequency depth component and the second frequency depth component is 1.

[0076] The frequency depth information is used to calculate and update the acquired component array, and after the calculation and update, the component array to be updated is generated for the next component update calculation.

[0077] After performing the component update calculation a preset number of times, the corresponding calculated and updated component array is configured as the component extraction target array.

[0078] Based on the component extraction target array, the in-phase component and the orthogonal component corresponding to each sampling point are determined.

[0079] As explained above, when performing echo acquisition processing to obtain ultrasonic echo signals, each ultrasonic echo signal includes multiple sampling points. The number of sampling points in the ultrasonic echo signal is related to the sampling frequency. The number of sampling points in the beamforming signal generated for the endoscopic echo should be consistent with the number of sampling points in the ultrasonic echo signal; that is, beamforming does not change the number of sampling points. When performing adaptive filtering, this invention extracts the sampling point components for at least each sampling point to generate in-phase and quadrature components corresponding to each sampling point. The details of the in-phase and quadrature components are explained below.

[0080] As explained above, since each sampling point needs to undergo sampling point component extraction processing once, when the endoscopic echo synthesized signal includes multiple sampling points, multiple sampling point component extraction processes should be performed to generate the in-phase and quadrature components corresponding to each sampling point. It should be understood that the sampling point represents the sampling position of the endoscopic echo synthesized signal. When performing sampling point component extraction processing, information such as the amplitude of the sampling point should also be utilized. The specific methods and processes for performing sampling point component extraction processing will be explained below.

[0081] In practice, the sampling point component extraction process should include a preset number of component update calculations. Figure 2 The diagram shows a flowchart of one embodiment of the present invention performing adaptive filtering. Figure 2 In this context, n represents the location of the sampling point, m represents the number of sampling points in each endoscopic echo synthesized signal, and numeter represents the preset number of iterations. In other words, the preset number of iterations numeter controls the number of loops for performing component update calculations. Specifically, the preset number of iterations numeter can be selected according to actual needs.

[0082] When performing component update calculation, the component array to be updated and the frequency depth information representing the depth state of the sampling point should be obtained. For each sampling point, the location of the sampling point remains unchanged during the sampling point extraction process, while the component array is updated after each component update calculation. Figure 2 In this context, P represents the component array to be updated, and F represents the frequency depth information.

[0083] In one embodiment of the present invention, for the frequency depth information of the current sampling point, the following is true:

[0084]

[0085] Where F represents the frequency depth information, s represents the first component of the frequency depth, c represents the second component of the frequency depth, f_mod represents the demodulation frequency, fs represents the echo sampling frequency, and n represents the sampling depth position of the current sampling point within the corresponding endoscopic echo synthesized signal.

[0086] Specifically, the demodulation frequency f_mod is consistent with the frequency of the excitation transmission process; that is, during Doppler detection, once the frequency of the excitation transmission process is determined, the demodulation frequency f_mod can be determined. As explained above, s 2 +c 2 =1. For each endoscopic echo synthesized signal, the sampling depth position n of the corresponding current sampling point within the corresponding endoscopic echo synthesized signal can be determined.

[0087] As can be seen from the above expression for frequency depth information, after calculating the first frequency depth component s and the second frequency depth component c based on the demodulation frequency f_mod, the frequency of the synthesized endoscopic echo signal can be shifted to near the zero intermediate frequency, thereby significantly reducing the complexity of subsequent signal processing and the requirements for hardware performance.

[0088] In practice, after obtaining the component array to be updated, the frequency depth information can be used to calculate and update the component array. After the calculation and update, a component array to be updated for the next component update calculation is generated, until the number of component update calculations reaches a preset number (numerator). Figure 2 As shown.

[0089] In one embodiment of the present invention, when the acquired component array is updated using frequency depth information, the following applies:

[0090]

[0091] Where F represents frequency depth information, and P is the component array to be updated. T Let F be the transpose of the frequency depth information, x(n) be the amplitude information of the current sampling point, mu be the convergence factor, and λ be the calculation update coefficient. This refers to the component array obtained after performing component update calculations. The dimension of the component array P to be updated is consistent with the dimension of the frequency depth information F.

[0092] Depend on Figure 2 It can be seen that when performing the sampling point component extraction process for each sampling point, the component array P should be initialized to P = [0,0]. Therefore, when performing the component update calculation process for the first time, the component array to be updated should be P = [0,0]. After performing the component update calculation process for the first time, the component array... This serves as the component array to be updated during the second component update calculation process, and so on for other cases.

[0093] Depend on Figure 2 As shown in the above expression, each time the component update calculation is performed, the component array to be updated should be transposed and multiplied by the frequency depth information to obtain the intermediate quantity H. As explained above, both the frequency depth information F and the component array P are 1x2 vectors. During the first component update calculation, the intermediate quantity H should be 0, the update coefficient λ can be 2, and the convergence factor mu should generally be a small value, such as 0.05. After the first component update calculation, the component array... If the value is not 0, then the component array can be... The value is assigned to the component array to be updated during the second component update calculation.

[0094] Each time the component update calculation is performed, the above calculation process can be repeated until the component update calculation is performed a preset number of times (numiter). Then, the component array obtained from the component update calculation when the preset number of times (numiter) is reached is used. The configuration is used as the target array for component extraction. Subsequently, based on the target array for component extraction, the in-phase component and the quadrature component corresponding to each sampling point are determined. Figure 2 The figure illustrates an embodiment of determining the in-phase and quadrature components corresponding to each sampling point based on the component extraction target array. The component array is shown in the figure. The first quantity As in-phase components, component array The second quantity As orthogonal components Figure 2 In this context, I(n) represents the in-phase component corresponding to the nth sampling point, and Q(n) represents the quadrature component corresponding to the nth sampling point. After extracting the in-phase and quadrature components corresponding to each sampling point, the nth sampling point can be expressed as: x(n) = I(n) + jQ(n).

[0095] As explained above, obtaining the in-phase and quadrature components corresponding to each sampling point enables adaptive filtering for that sampling point. As described above regarding adaptive filtering, the adaptive filtering of this invention is based on the characteristics of ultrasound echo signals and noise reflected from the digestive tract. After component update calculations with a preset number of numeric steps, the error between the filtered endoscopic signal and the desired blood flow signal can be minimized. Furthermore, the preset number of numeric steps and the convergence factor mu can be dynamically adjusted to further improve the adaptive filtering effect. The details of dynamically adjusting the preset number of numeric steps and the convergence factor mu are provided below.

[0096] Depend on Figure 1 It is known that each endoscopic echo synthesis signal, after adaptive filtering, can generate a corresponding endoscopic synthesis filtered signal. Based on all the endoscopic synthesis filtered signals, an endoscopic synthesis filtered signal group can be formed. Subsequently, spectral estimation based on feature representation can be performed on the endoscopic synthesis filtered signal group to generate the Doppler spectrum signal of the current Doppler ultrasound detection. It should be understood that the obtained Doppler spectrum signal is the signal after blood flow noise suppression in this invention. The method and process of spectral estimation of the endoscopic synthesis filtered signal group are described in detail below.

[0097] In one embodiment of the present invention, spectral estimation based on feature representation includes:

[0098] Based on the endoscopic synthesized and filtered signal group, the corresponding autocorrelation matrix R is constructed, and the noise subspace matrix G of the autocorrelation matrix R is calculated.

[0099] Construct a steering vector with normalized frequency, and calculate the corresponding Doppler spectrum signal of the current endoscopic synthesized and filtered signal group based on the constructed steering vector and the noise subspace matrix G.

[0100] Specifically, for the constructed autocorrelation matrix R, we have:

[0101]

[0102] Where N is the number of endoscopic synthesized and filtered signals in the endoscopic synthesized and filtered signal group, L is the number of valid samples, and M is the number of observation samples set, L = N - M + 1, [x1(i)] HT Let x1(i) be the conjugate transpose of x1(i), S(i) be the synthesized and filtered signal of the i-th endoscope, and S(i-1) be the synthesized and filtered signal of the (i-1)-th endoscope.

[0103] In practical implementation, when constructing the autocorrelation matrix R, the number of observation samples M should be configured. The number of observation samples M can generally be selected according to actual needs. Generally, when selecting the number of observation samples M, it should be ensured that N≥10M to avoid estimation bias of the autocorrelation matrix R. For example, the number of observation samples M can be 4. Of course, the number of observation samples M can also be other values, which can be selected according to needs. Examples will not be given here. After determining the number of observation samples M, the number of elements in x1(i) can be determined. Specifically, the number of elements in x1(i) is M. For example, when M is 4, the corresponding x1(4) is: x1(4)=[S(4),S(3),S(2),S(1)]. Other cases can be referred to here. Examples will not be given here. After determining x1(i), the conjugate transpose of x1(i) [x1(i)] can be determined accordingly. H .

[0104] It should be understood that since each sampling point can be represented as in-phase and quadrature components, the i-th endoscopic synthesized and filtered signal S(i) should also be composed of the corresponding in-phase and quadrature components. In this case, x1(i)[x1(i)] is executed. HT During the calculation, the commonly used calculations for complex numbers can be adopted. The specific calculations are consistent with the existing technology and will not be elaborated here.

[0105] As explained above, the i-th endoscopic synthesized and filtered signal S(i) can be represented by the combination of the corresponding in-phase and quadrature components. Therefore, for a given set of endoscopic synthesized and filtered signals, the corresponding autocorrelation matrix R can be constructed. After constructing the autocorrelation matrix R, the noise subspace matrix G of the autocorrelation matrix R can be calculated. The method and process for calculating the noise subspace matrix G are explained in detail below.

[0106] In one embodiment of the present invention, when calculating the noise subspace matrix G of the autocorrelation matrix R, the following is obtained:

[0107] The autocorrelation matrix R is subjected to eigenvalue decomposition to obtain the eigenvalues ​​of the autocorrelation matrix R and the eigenvector corresponding to each eigenvalue;

[0108] Sort the eigenvalues ​​of the autocorrelation matrix R in descending order, select the eigenvector components corresponding to the (k+1)th eigenvalue to the Mth eigenvalue, and generate the noise subspace matrix G based on the selected eigenvector components.

[0109] Specifically, when calculating the noise subspace matrix G, the autocorrelation matrix R should first be eigenvalued. For example, the autocorrelation matrix can be expressed as: R = VΦV T Where Φ is the eigenvalue diagonal matrix, V is the eigenvector, and V T As the transpose of the eigenvectors, the eigenvalue decomposition of the autocorrelation matrix R can be performed in the same way as before, thus obtaining the eigenvalue diagonal matrix Φ and the eigenvector V. It should be noted that after eigenvalue decomposition of the autocorrelation matrix R, M eigenvalues ​​are obtained, that is, the eigenvalue diagonal matrix Φ is an M*M square matrix.

[0110] After obtaining all eigenvalues ​​of the autocorrelation matrix R using the above method, the eigenvalues ​​can be sorted in descending order. The first k eigenvalues ​​correspond to the signal subspace, and the (k+1)th to Mth eigenvalues ​​correspond to the noise subspace. Therefore, the noise subspace matrix G can be constructed based on the eigenvector components corresponding to the (k+1)th to Mth eigenvalues. Specifically:

[0111] Generally, k can be taken as about 3 / 4 of M. Therefore, after selecting the number of observation samples M, the value of k can be determined accordingly. Of course, the value of k can also be selected according to the actual application, specifically to effectively achieve noise suppression. In other words, the value of k can be finely adjusted according to blood flow noise suppression.

[0112] It should be noted that the Doppler spectrum signal is a signal that varies with frequency. Therefore, in order to perform spectrum estimation, a steering vector with normalized frequency should be constructed. In one embodiment of the present invention, when constructing the steering vector with normalized frequency, the following is done:

[0113]

[0114] Among them, f g For the normalized frequency, the normalized frequency f g The value range is 0 to 1, and the normalized frequency f g The step length is 1 / M;

[0115] Specifically, the normalized frequency f g The quantity is a gradually increasing value from 0 to 1, and each increment should be 1 / M. Therefore, a corresponding normalized frequency f is determined. g Then, a corresponding guiding vector can be obtained, which contains M elements.

[0116] In practical implementation, the Doppler spectrum signal corresponding to the current endoscopic synthesized and filtered signal group is calculated based on the constructed steering vector and the noise subspace matrix G. Specifically, when calculating the generated Doppler spectrum signal, the following applies:

[0117]

[0118] Where y(f) is the Doppler spectral signal, G T Let be the transpose of the noise subspace matrix G, (aw) T It is the transpose of the guiding vector.

[0119] As explained above, the spatial dimension of the noise subspace matrix G is (Mk), and this dimension is typically smaller than the dimension k of the signal subspace. Therefore, calculating the Doppler signal based on the noise subspace matrix G is more efficient. Furthermore, the noise subspace matrix G provides a more direct representation of blood flow noise distribution. When calculating the Doppler signal using this method, orthogonality detection between the noise subspace and the steering vector can be achieved. This means the frequency of the Doppler spectrum signal can guarantee the frequency for orthogonality detection, better suppressing noise influence and resulting in sharper peaks at the true frequency, leading to higher resolution. Therefore, compared to using the signal subspace, using the noise subspace matrix G and the steering vector to calculate the Doppler signal not only improves computational efficiency but also enhances the accuracy and reliability of the calculated Doppler spectrum signal.

[0120] It should be understood that y(f) mentioned above is the spectral function of the Doppler signal. When actually plotting the image, a logarithmic approach can be taken, resulting in: Y(f) = 10lg(y(f)). Furthermore, frequency estimation can be performed on the graphical correlation of Y(f). Frequency estimation involves finding the peak position of Y(f), and the frequency corresponding to the peak is the estimated frequency value.

[0121] In practical implementation, after obtaining the frequency estimate, spectral quality assessment can be performed based on the frequency estimate. During spectral quality assessment, the ratio of the peak position of Y(f) to the average noise during Doppler ultrasound detection can be calculated; that is, the ratio can be used for spectral quality assessment. Based on the spectral quality assessment status, the aforementioned dynamically adjusted preset number of iterations (numitter) and convergence factor (mu) can be adjusted until the spectral quality is optimal. This allows for the dynamic adjustment of the preset number of iterations (numitter) and convergence factor (mu). Of course, other methods can also be used to achieve the dynamic adjustment of the preset number of iterations (numitter) and convergence factor (mu), which will not be illustrated here.

[0122] As explained above, noise is usually random during Doppler imaging of the digestive tract. When generating the Doppler spectrum signal, the aforementioned adaptive filtering and autocorrelation calculations effectively distinguish noise from blood flow signals. This means that by utilizing the characteristic differences between noise and blood flow signals, they can be precisely separated, thus effectively suppressing blood flow noise. This solves the problem of blood flow noise in digestive tract endoscopic imaging in this technical field. Therefore, during digestive tract endoscopic imaging, dynamic real-time optimization of endoscopic imaging can be achieved, eliminating the need for medical personnel to manually adjust the image later, improving diagnostic efficiency by approximately 50%, and making it particularly suitable for emergency endoscopic examinations.

[0123] Based on the above description, a Doppler blood flow noise suppression system suitable for gastrointestinal endoscopic examination can be obtained. Specifically, it includes an endoscopic detection device and an endoscopic processor adapted and connected to the endoscopic detection device, wherein...

[0124] When performing Doppler ultrasound examination on the digestive tract, the endoscope processor and the endoscope detection device use the Doppler blood flow noise suppression method described above to suppress noise.

[0125] Figure 3 The diagram illustrates a structural block diagram of an embodiment of a Doppler blood flow noise suppression system. As shown in the diagram, the Doppler blood flow noise suppression system may include an endoscopic detection device and an endoscopic processor. The endoscopic detection device can be described above. The endoscopic processor can be a commonly used processing device, such as a computer. The type of endoscopic processor can be selected as needed. When the endoscopic processor works in conjunction with the endoscopic detection device, it can control the endoscopic detection device to perform the above-mentioned excitation transmission processing and echo acquisition processing. The endoscopic processor can perform the above-mentioned beamforming processing, or perform the above-mentioned adaptive filtering and spectral estimation, etc. The specific selection can be made as needed, based on the ability to achieve the Doppler blood flow noise suppression.

Claims

1. A Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination, characterized in that, The Doppler blood flow noise suppression method includes: An endoscopic detection device is provided, and the endoscopic detection device is configured to perform Doppler ultrasound detection on the digestive tract to be examined in an endoscopic manner, so as to generate an endoscopic scan echo signal group after Doppler ultrasound detection, wherein the endoscopic scan echo signal group includes several endoscopic echo composite signals. Adaptive filtering is applied to the endoscopic echo synthesized signals within the endoscopic scan echo signal group, and an endoscopic synthesized filtered signal is generated after adaptive filtering. Furthermore, an endoscopic synthesized filtered signal group is generated based on all the endoscopic synthesized filtered signals. When adaptively filtering each endoscopic echo synthesized signal, at least the in-phase component and quadrature component corresponding to each sampling point in the current endoscopic echo synthesized signal must be determined. The above-mentioned endoscopic synthesized and filtered signal group is subjected to spectral estimation based on feature representation, so as to generate the Doppler spectrum signal of the current Doppler ultrasound detection after spectral estimation; When performing adaptive filtering on the synthesized endoscopic echo signal, at least one sampling point component extraction process is performed on each sampling point within the synthesized endoscopic echo signal. This extraction process generates in-phase and quadrature components corresponding to each sampling point. When performing sample point component extraction processing, it includes performing component update calculation processing a preset number of times, wherein, When performing component update calculations, the following steps are included: At least the component array to be updated and the frequency depth information representing the current sampling point depth state are obtained, wherein the frequency depth information includes a first frequency depth component and a second frequency depth component, and the sum of squares of the first frequency depth component and the second frequency depth component is 1. The frequency depth information is used to calculate and update the acquired component array, and after the calculation and update, the component array to be updated is generated for the next component update calculation. After performing component update calculations a preset number of times, the corresponding calculated and updated component array is configured as the component extraction target array. Based on the component extraction target array, the in-phase component and the orthogonal component corresponding to each sampling point are determined.

2. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to claim 1, characterized in that, When using frequency depth information to calculate and update the acquired component array, we have: in, For frequency depth information, For the component array to be updated, For frequency depth information transpose, This provides the amplitude information for the current sampling point. The convergence factor is To calculate the update coefficients, This refers to the component array obtained after performing component update calculations. Component array to be updated Corresponding dimensional and frequency depth information The corresponding dimensions are consistent.

3. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to claim 2, characterized in that, For the frequency depth information of the current sampling point, we have: in, For frequency depth information, The first component of the frequency depth. The second component of frequency depth, For demodulation frequency, The echo sampling frequency, This represents the sampling depth position of the current sampling point within the corresponding endoscopic echo synthesized signal.

4. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to any one of claims 1 to 3, characterized in that, When performing spectral estimation based on feature representations, the following steps are included: Based on the endoscopic synthesized and filtered signal group, the corresponding autocorrelation matrix is ​​constructed. And calculate the autocorrelation matrix. noise subspace matrix ; Construct a normalized frequency steering vector, and based on the constructed steering vector and the noise subspace matrix... Calculate the corresponding Doppler spectrum signal of the current endoscopic synthesized and filtered signal group.

5. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to claim 4, characterized in that, For the constructed autocorrelation matrix Then we have: in, This represents the number of endoscopic synthesized and filtered signals within the endoscopic synthesized and filtered signal group. For the number of valid samples, To set the number of observation samples, , for The conjugate transpose of . For the first The signal is synthesized and filtered by an endoscope. For the first The signal is synthesized and filtered by an endoscope.

6. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to claim 5, characterized in that, Calculate the autocorrelation matrix noise subspace matrix Then: For autocorrelation matrix Perform eigenvalue decomposition to obtain the autocorrelation matrix. The eigenvalues ​​and the eigenvectors corresponding to each eigenvalue; autocorrelation matrix Sort the eigenvalues ​​from largest to smallest, and select the (k+1)th eigenvalue to the ... The eigenvector components corresponding to each eigenvalue are used to generate a noise subspace matrix based on the selected eigenvector components. , where k takes the value of 3 / 4 of the quantity M.

7. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to claim 5, characterized in that, When constructing the steering vector of the normalized frequency, we have: in, Normalized frequency, normalized frequency The value range is 0~1, and the normalized frequency is... The step length is 1 / M; When calculating the generated Doppler spectrum signal, we have: in, The signal is a Doppler spectrum. The noise subspace matrix The transpose of the matrix, It is the transpose of the guiding vector.

8. The Doppler blood flow noise suppression method suitable for gastrointestinal endoscopic examination according to any one of claims 1 to 3, characterized in that, When performing Doppler ultrasound examination on the digestive tract, it includes several sequential ultrasound scans. Each ultrasound scan includes sequential excitation emission processing, echo acquisition processing, and beamforming processing. During the excitation transmission process, each array element in the endoscopic detection device transmits an ultrasonic signal into the digestive tract to be examined. During echo acquisition and processing, each array element in the endoscopic detection device is configured to receive the ultrasound echo signal reflected from the digestive tract to be examined; When performing beamforming, the ultrasonic echo signals received by all array elements are beamformed to generate a single endoscopic echo composite signal. The number of synthesized endoscopic echo signals in the endoscopic scan echo signal group is consistent with the number of times the excitation emission process is performed.

9. A Doppler blood flow noise suppression system suitable for endoscopic examination of the digestive tract, characterized in that, It includes an endoscopic detection device and an endoscopic processor adapted and connected to the endoscopic detection device, wherein, When performing Doppler ultrasound examination on the digestive tract, the endoscope processor and the endoscope detection device use the Doppler blood flow noise suppression method described in any one of claims 1 to 8 to suppress noise.

Images