An ultrasonic radio frequency signal quantitative parameter estimation method, device, medium and product

Abstract

The application discloses an ultrasonic radio frequency signal quantitative parameter estimation method, device, medium and product, relates to the technical field of ultrasonic imaging and signal processing, and comprises the following steps: acquiring an ultrasonic radio frequency signal reflected by a to-be-detected tissue and performing pretreatment to obtain a normalized RF signal and construct a joint objective function; the joint objective function integrates a data fitting term, an L2 norm regular constraint term of a backscattering coefficient, an L2 norm regular constraint term of an attenuation coefficient and a joint TV norm space smoothing constraint term; the constraint coefficients of the joint objective function are determined based on the relative signal-to-noise ratio of the normalized RF signal and the tissue attribute of the to-be-detected tissue; all the constraint coefficients satisfy a physical constraint condition greater than 0; the quantitative estimation value of the quantitative parameter of the to-be-detected tissue is estimated based on the joint objective function and all the constraint coefficients; and a corresponding spatial distribution image is generated according to the quantitative estimation value, so that the application meets the high-precision requirement of clinical actual application and has high universality.

Description

Technical Field

[0001] This application relates to the field of ultrasound imaging and signal processing technology, and in particular to a method, device, medium and product for estimating quantitative parameters of ultrasound radio frequency signals. Background Technology

[0002] Quantitative ultrasound technology is the core direction of the development of ultrasound imaging from qualitative diagnosis to quantitative assessment. Its core lies in estimating the backscattered scattering coefficient (BSC) and attenuation coefficient in the ultrasound radio frequency (RF) signal. This allows for the quantitative characterization of the microstructure and physical properties of the tested medium (such as human tissue). Among these, the BSC reflects the density, size distribution, and acoustic properties of the scatterers in the medium. Reflecting the energy attenuation characteristics of ultrasound waves propagating in a medium, both ultrasound reference phantoms and ultrasound reference structures have irreplaceable application value in the medical field (such as differentiating benign and malignant tumors, staging liver fibrosis, and assessing cardiovascular diseases). They can be compared using ultrasound reference phantoms to compare the effects of ultrasound on the BSC (Body Strain Detection System). The estimation accuracy was verified.

[0003] Existing quantitative parameter estimation methods for ultrasound RF signals mainly include the least squares method, maximum likelihood estimation, spectral difference method, and spectral shift method. These methods have the following technical shortcomings, making it difficult to meet the high accuracy and robustness requirements of practical applications:

[0004] 1. Poor robustness: Ultrasonic RF signals are easily affected by factors such as noise, probe consistency, and propagation attenuation. In low signal-to-noise ratio (SNR) scenarios (such as deep tissue imaging), the estimation error increases sharply and even parameter distortion occurs. The estimation stability in such scenarios can be verified by using a reference phantom.

[0005] 2. Inappropriate constraint settings: BSC and attenuation coefficient were not simultaneously considered. Setting independent L2 norm constraints without setting joint TV norm constraints for both makes it impossible to effectively balance parameter stability and joint smoothness, resulting in estimation results that are prone to extreme values ​​or local abrupt changes.

[0006] 3. Unscientific determination of constraint coefficients: Constraint coefficients are often set manually without being adjusted according to the data SNR and organizational attributes. Furthermore, the minimum value of the coefficients is not standardized to be greater than 0, which can easily lead to constraint failure or over-constraint, affecting the estimation accuracy.

[0007] 4. Low solution efficiency: BSC and The joint estimation problem is a non-convex, ill-conditioned optimization problem. Existing solution algorithms have slow convergence speed and high computational complexity, making it difficult to meet the needs of real-time imaging and detection.

[0008] 5. Incomplete parameter decoupling: The relationship between BSC and... The coupling relationship between the two is not effectively compensated for, and the estimation accuracy of the two is affected by each other, which further reduces the reliability of quantitative assessment.

[0009] In recent years, some studies have employed the Alternating Direction Method of Multipliers (ADMM) to solve ultrasound parameter estimation problems in an attempt to improve solution efficiency. However, this type of method still suffers from problems such as unreasonable constraint terms, unscientific coefficient determination, and the use of fixed regularization coefficients. The selection of regularization coefficients lacks theoretical basis, fails to incorporate SNR and tissue properties for adjustment, and does not standardize the minimum value of the coefficients, resulting in insufficient estimation accuracy and robustness. Accuracy verification largely relies on reference phantoms, but even with reference phantom verification, existing methods still struggle to meet the high-precision requirements of practical clinical applications. Furthermore, existing methods do not establish a correlation between constraint coefficients and the inherent characteristics of RF signals (such as SNR) and tissue properties, leading to poor versatility and difficulty in adapting to the diverse needs of the medical field. Summary of the Invention

[0010] The purpose of this application is to provide a method, device, medium, and product for estimating quantitative parameters of ultrasound radiofrequency signals, in order to solve the problems of difficulty in meeting the high precision requirements of clinical applications and poor versatility.

[0011] To achieve the above objectives, this application provides the following solution.

[0012] In a first aspect, this application provides a method for estimating quantitative parameters of ultrasonic radio frequency signals, comprising the following steps.

[0013] The ultrasound radio frequency signal reflected from the tissue under test is acquired and preprocessed to obtain a normalized RF signal; the preprocessing includes beamforming, noise reduction and normalization.

[0014] The normalized RF signal is analyzed using a sliding window approach, and a joint objective function is constructed for the normalized RF signal within each sliding window. The joint objective function integrates a data fitting term, an L2 norm regularization constraint term for the backscattering coefficient, an L2 norm regularization constraint term for the attenuation coefficient, and a joint TV norm spatial smoothing constraint term for the backscattering coefficient and the attenuation coefficient.

[0015] The constraint coefficients of the joint objective function are determined based on the relative signal-to-noise ratio of the normalized RF signal and the tissue properties of the tissue under test; all constraint coefficients satisfy physical constraint conditions greater than 0.

[0016] Based on the joint objective function and all constraint coefficients, quantitative estimates of the quantitative parameters of the tissue under test are estimated; the quantitative parameters include backscattering coefficient and attenuation coefficient.

[0017] Based on the quantitative estimate, a corresponding spatial distribution image is generated; the spatial distribution image is used to analyze the properties of the tissue to be tested.

[0018] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for estimating quantitative parameters of ultrasonic radio frequency signals.

[0019] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for estimating quantitative parameters of ultrasonic radio frequency signals.

[0020] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for estimating quantitative parameters of ultrasonic radio frequency signals.

[0021] According to the specific embodiments provided in this application, this application has the following technical effects: This application constructs a joint objective function integrating an integrated data fitting term, an L2 norm regularization constraint term for the backscattering coefficient, an L2 norm regularization constraint term for the attenuation coefficient, and a joint TV norm spatial smoothing constraint term for the backscattering coefficient and the attenuation coefficient. Based on the relative SNR of the normalized RF signal and the tissue properties of the tissue under test, the constraint coefficients of the joint objective function are determined, and all constraint coefficients are guaranteed to satisfy a physical constraint condition greater than 0, thus realizing the integration of BSC and... It provides high-precision and robust joint estimation to generate corresponding spatial distribution images for analyzing the properties of the tissue under test. It is more suitable for medical ultrasound scenarios and can meet the high-precision requirements of clinical applications. It does not require additional hardware modifications, is compatible with existing ultrasound imaging systems, and has high versatility. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a quantitative parameter estimation method for ultrasonic radio frequency signals provided in an embodiment of this application.

[0024] Figure 2 This is a flowchart illustrating another method for estimating quantitative parameters of ultrasound radio frequency signals for medical ultrasound tissue characterization, provided as an embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] To make the objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] In this application, "ultrasound probe", "RF signal", "BSC" and " Terms such as "ADMM algorithm," "L2 norm," and "TV norm" are common terms in the field of ultrasound signal processing, and their meanings are consistent with the conventional understanding in this field, requiring no further explanation. The parameters in this application (such as...) , , , Empirical values ​​(etc.) can be fine-tuned according to actual medical application scenarios, probe parameters, and the tissue to be tested, and are all within the scope of protection of this application; and all constraint coefficients (etc.) , , All values ​​must be strictly greater than 0 to ensure that the constraints are effective.

[0028] like Figure 1 As shown in the figure, this application provides a method for estimating quantitative parameters of ultrasonic radio frequency signals, including the following steps.

[0029] S1: Acquire the ultrasound radio frequency signal reflected back from the tissue under test and perform preprocessing to obtain a normalized RF signal; the preprocessing includes beamforming, noise reduction and normalization.

[0030] S2: The normalized RF signal is analyzed using a sliding window method, and a joint objective function is constructed for the normalized RF signal within each sliding window; the joint objective function integrates the data fitting term, the L2 norm regularization constraint term of the backscattering coefficient, the L2 norm regularization constraint term of the attenuation coefficient, and the joint TV norm spatial smoothing constraint term of the backscattering coefficient and the attenuation coefficient.

[0031] S3: Determine the constraint coefficients of the joint objective function based on the relative signal-to-noise ratio of the normalized RF signal and the tissue properties of the tissue under test; all constraint coefficients satisfy the physical constraint condition of being greater than 0.

[0032] S4: Based on the joint objective function and all constraint coefficients, estimate the quantitative parameters of the tissue under test; the quantitative parameters include the backscattering coefficient and the attenuation coefficient.

[0033] S5: Generate a corresponding spatial distribution image based on the quantitative estimate; the spatial distribution image is used to analyze the properties of the tissue to be tested.

[0034] This application is applicable to medical ultrasound tissue characterization scenarios and can achieve backscatter coefficient (BSC) and attenuation coefficient (BSC). High-precision and robust joint estimation.

[0035] In an exemplary embodiment, S1 specifically includes: first, emitting ultrasonic waves through an ultrasonic probe and receiving RF signals reflected by the tissue to be tested. The dimension of the RF signal is [number of depth sampling points D, number of ultrasonic probe elements N], where the number of depth sampling points D corresponds to the depth of ultrasonic wave propagation, and the number of ultrasonic probe elements N corresponds to the number of ultrasonic probe elements used.

[0036] The received RF signals are subjected to the following preprocessing operations in sequence.

[0037] (1) Beamforming: The signals received by each element of the ultrasonic probe are delayed and weighted to focus on each depth point of the tissue to be tested, thereby improving the spatial resolution of the signal.

[0038] (2) Denoising: Gaussian filtering or wavelet denoising algorithm is used to filter out high-frequency noise and random interference in the RF signal, retain the effective signal components, and avoid the influence of noise on subsequent parameter estimation.

[0039] (3) Normalization: Normalization is achieved by dividing the spectrum of the tissue under test by the spectrum of a standard phantom. The standard phantom is an ultrasonic reference phantom with known acoustic characteristics, which can effectively eliminate interference from system parameters such as ultrasonic probe frequency response and system gain, and normalize the RF signal amplitude to the same order of magnitude, ensuring the normalized RF signal. The amplitude is only related to the characteristics of the tissue under test and is independent of the frequency, providing a stable basis for subsequent parameter estimation.

[0040] In an exemplary embodiment, S2 specifically includes: constructing a joint objective function using the backscattering coefficient and attenuation coefficient corresponding to the normalized RF signal within each sliding window as optimization variables.

[0041] In an exemplary embodiment, the normalized RF signal is analyzed using a sliding window method, specifically including: selecting a rectangular window with a side length that is an integer multiple of the wavelength of the normalized RF signal as a sliding window, and sliding it in a zigzag manner over the entire target ultrasonic radio frequency signal with a preset step size to analyze the normalized RF signal; the zigzag manner is that the sliding window first slides along a first direction to the end, then moves one step in the vertical direction, and then slides in the opposite direction along the first direction, and so on until the entire target ultrasonic radio frequency signal area is covered.

[0042] In practical applications, the normalized RF signal is analyzed using a sliding window approach: a rectangular window of a certain side length is selected (the window side length can be set to an integer multiple of the ultrasound RF signal wavelength, such as 10 times the wavelength). The window slides across the entire target ultrasound RF signal in a specific step size (e.g., 1 / 4 of the window side length, ensuring some overlap between windows), following a certain sequence (e.g., a zigzag sliding method, specifically as follows: the window first slides along one direction of the ultrasound RF signal (e.g., the horizontal direction) to the end of that direction, then moves one step vertically, and then slides back horizontally, repeating this process to form a zigzag-like sliding trajectory. This ensures that the entire target ultrasound RF signal area is covered by the window, avoids redundant calculations caused by repeated scanning, and guarantees the uniformity of the scan). For the normalized RF signal within each sliding window, the backscattering coefficient and attenuation coefficient of the tissue it represents are calculated.

[0043] In practical applications, the data fitting term is used to measure the fitting error between the normalized RF signal and the model prediction value, wherein the model prediction value includes the attenuation factor of the ultrasound at the corresponding depth; the L2 norm regularization constraint term of the attenuation coefficient is used to constrain the amplitude of the attenuation coefficient; the L2 norm regularization constraint term of the backscattering coefficient is used to constrain the amplitude of the backscattering coefficient; the joint TV norm spatial smoothness constraint term, which includes the joint gradient of the attenuation coefficient and the backscattering coefficient, is used to constrain the joint spatial smoothness of the attenuation coefficient and the backscattering coefficient.

[0044] Specifically, the backscattering coefficients corresponding to the normalized RF signal within the sliding window. and attenuation coefficient To optimize the variables, a joint objective function is constructed. The joint objective function integrates the data fitting term, the L2 norm regularization constraint term of the BSC, and the attenuation coefficient. L2 norm regularization constraints and BSC and The joint TV norm spatial smoothing constraint term, which balances estimation accuracy, parameter stability, and joint smoothness, is specifically expressed as follows:

[0045]

[0046] The functions of each component are as follows:

[0047] (1) Data fitting term: Based on the attenuation and backscattering model of ultrasonic signal propagation, the normalized RF signal is measured. Compared with model predictions ( The fitting error of ), where, For ultrasound at depth The attenuation factor at the location ensures that the estimated results closely match the measured data;

[0048] (2) Attenuation coefficient L2 norm regularization constraint terms: , Attenuation coefficient The constraint coefficients of the L2 norm regularization constraint term ( The attenuation coefficient is determined by the properties of the tissue being tested and is used to constrain the attenuation coefficient. The amplitude, to prevent overfitting and ensure To ensure continuity and stability, and to avoid extreme values ​​in the estimation results.

[0049] (3) L2 norm regularization constraint terms of BSC: , The constraint coefficients of the L2 norm regularization constraint term for the backscattering coefficients ( The backscattering coefficient (BSC) is determined by the properties of the tissue under test and is used to constrain the amplitude of the BSC, prevent abnormal fluctuations in the BSC estimation, and ensure the stability and rationality of the BSC.

[0050] (4) BSC and Joint TV norm space smoothness constraint term: , The constraint coefficients for the joint TV norm space smoothing constraint term ( ), determined by the relative SNR of the data, for The joint gradient with BSC is used for constraints. The joint spatial smoothness with BSC avoids local abrupt changes in both, while also taking into account their coupling relationship, thus improving the accuracy of joint estimation.

[0051] All constraint coefficients ( , , All values ​​are strictly greater than 0, and the minimum value is greater than 0 to ensure that the constraints are effective and to avoid the problems of constraint failure (coefficient is 0) or negative constraints (coefficient is less than 0).

[0052] S3 specifically includes: determining the constraint coefficients of the joint TV norm spatial smoothing constraint term based on the relative signal-to-noise ratio of the normalized RF signal; and determining the constraint coefficients of the L2 norm regularization constraint term of the attenuation coefficient and the L2 norm regularization constraint term of the backscattering coefficient based on the tissue properties of the tissue under test.

[0053] In practical applications, the constraint coefficients of the joint TV norm spatial smoothing constraint term are determined based on the relative signal-to-noise ratio of the normalized RF signal, specifically including:

[0054] The region of unorganized reflection signal in the normalized RF signal is selected as the noise window, and the noise power at each frequency point is calculated.

[0055] Based on the noise power, the relative signal-to-noise ratio at each depth-frequency point is calculated, and the relative signal-to-noise ratio is limited to a preset range.

[0056] The relative signal-to-noise ratio within the preset range is averaged along the depth dimension to obtain the frequency-specific relative signal-to-noise ratio.

[0057] The constraint coefficients of the joint TV norm spatial smoothing constraint term are calculated by linear mapping based on the relative signal-to-noise ratio per frequency.

[0058] Specifically, Based on adaptive adjustment of relative SNR of data and Based on the physical properties of the tested tissue, and with all coefficients having a minimum value greater than 0, the specific steps are as follows.

[0059] (1) Noise power calculation: Select the region of no tissue reflection signal in the normalized RF signal (such as the ultrasound far-field blank region, which has no tissue reflection signal and only contains background noise) as the noise window, and calculate the noise power at each frequency point. To avoid division by zero error, The lower limit is ;in, This refers to the spectrum of the noise signal at frequency point f within the noise window (noise_window). For frequency index, F represents the total number of frequencies.

[0060] (2) Calculation of relative signal-to-noise ratio: For each depth-frequency point ( ), calculate the relative signal-to-noise ratio ,in, Let be the signal power at that point; to avoid extreme values ​​affecting subsequent coefficient calculations, Limited to Within the range, This represents the maximum signal-to-noise ratio. 40dB (which is within the normal SNR range of ultrasound signals).

[0061] (3) Frequency average: for The relative signal-to-noise ratio is obtained by averaging along the depth dimension. This simplifies coefficient calculation while ensuring frequency adaptability of the coefficients.

[0062] (4) Determination of coefficients: ① :based on The core principle behind linear mapping calculations is that in high SNR scenarios, the signal-to-noise ratio is low, thus reducing... To avoid excessive smoothing that could lead to loss of parameter details; in low SNR scenarios, noise accounts for a high proportion, increasing... Strengthen joint smoothing constraints to suppress noise while ensuring The specific formula is as follows:

[0063]

[0064] in, for The minimum value (empirical value is 0.01, strictly greater than 0). for The maximum value (empirical value taken as 0.5) ensures It always stays within the range of (0, 0.5], ensuring that the constraint is effective and not excessive;

[0065] ② Based on the physical properties of the tested tissue, different human tissues (such as liver, kidney, and tumor tissue) have different acoustic characteristics (density, scatterer distribution, attenuation characteristics), and corresponding... The values ​​are different, and all are strictly greater than 0, for example, the values ​​corresponding to liver tissue. The value ranges from 0.02 to 0.08 to ensure that the constraint strength matches the organizational characteristics.

[0066] ③ Also determined based on the attributes of the organization being tested, and... Synchronous adaptation to tissue characteristics, all of which are strictly greater than 0, such as the characteristics corresponding to liver tissue. The value range is 0.05~0.15 to avoid abnormal fluctuations in BSC estimation, and at the same time... Synergistic effects enhance the stability of joint estimation.

[0067] S4 specifically includes: based on the joint objective function and all constraint coefficients, introducing auxiliary variables, decomposing and optimizing the variables using the alternating direction multiplier method, solving the sub-problems through alternating iterations and performing convergence determination, and obtaining quantitative estimates of the quantitative parameters of the tissue under test; the sub-problems include the joint sub-problem of attenuation coefficient and backscattering coefficient and the auxiliary variable sub-problem; the quantitative estimates include the quantitative estimates of attenuation coefficient and backscattering coefficient.

[0068] In an exemplary embodiment, based on the joint objective function and all constraint coefficients, auxiliary variables are introduced, and the optimization variables are decomposed using the alternating direction multiplier method. By iteratively solving subproblems and determining convergence, quantitative estimates of the quantitative parameters of the tissue under test are obtained, specifically including:

[0069] By introducing auxiliary variables corresponding to the joint TV norm space smoothing constraint term, the joint objective function is transformed into an optimization problem with equality constraints.

[0070] Based on the optimization problem with equality constraints, dual variables and penalty parameters are introduced to construct an augmented Lagrangian function.

[0071] Based on the augmented Lagrangian function, the joint subproblem of attenuation coefficient and backscattering coefficient and the auxiliary variable subproblem are solved sequentially, and the dual variable is updated; wherein, the joint subproblem of attenuation coefficient and backscattering coefficient is solved by the conjugate gradient method or the L-BFGS-B quasi-Newton method, and the auxiliary variable subproblem is solved by the TV norm Chambolle projection operator.

[0072] Set a convergence threshold and a maximum number of iterations. Stop iterating when the iterative change of the optimization variable is less than the convergence threshold or the maximum number of iterations is reached, and output the quantitative estimate of the quantitative parameters of the tissue to be tested.

[0073] In practical applications, the ADMM algorithm, by decomposing optimization variables and iteratively solving subproblems, can efficiently solve non-convex and ill-conditioned optimization problems. Compared with traditional algorithms, it has a faster convergence speed, lower computational complexity, and can quickly obtain BSC and The estimation results are used to verify the estimation accuracy by referring to the phantom. The specific steps are as follows:

[0074] (1) Variable splitting: introducing auxiliary variables The joint objective function is transformed into an optimization problem with equality constraints, corresponding to the joint TV norm space smoothing constraint, thus reducing the difficulty of solving the problem.

[0075]

[0076] in, This represents the minimization operation, and st represents the constraint condition. The gradient is the combined gradient of the attenuation coefficient and the backscattering coefficient.

[0077] (2) Construction of augmented Lagrangian function: introduction of dual variables With penalty parameters In this embodiment, If the expression is greater than 0, the equality constraints are incorporated into the joint objective function to construct the augmented Lagrangian function. :

[0078]

[0079] in, Indicates the inner product, penalty parameter Used to adjust the strength of constraints and accelerate iterative convergence.

[0080] (3) Alternating iterative update: Fix some variables and solve them sequentially. Combine the subproblems with BSC and u-subproblems, and update the dual variables. The details are as follows:

[0081] ① Joint subproblem with BSC: Fixed , Solve about The minimum value of BSC, that is: ,in, The decay coefficient is the value updated after the (k+1)th iteration. The backscattering coefficients are those updated after the (k+1)th iteration. The gradient of the decay and backscattering coefficients after the k-th iteration update (i.e., auxiliary variable) is the joint gradient of the decay and backscattering coefficients. Let be the dual variable after the k-th iteration update.

[0082] This subproblem is a convex optimization problem. Solving it using the conjugate gradient method or the L-BFGS-B quasi-Newton method can quickly converge to the optimal solution, reducing computational complexity while also considering... The coupling relationship with BSC improves the accuracy of joint estimation.

[0083] ② Subproblem u: Fixed , , Solve about The minimum value of this subproblem is found, and the analytical solution is the TV norm Chambolle projection operator. The specific implementation steps are as follows: Calculate the norm and normalize it, then obtain it through projection operation. Update value To ensure that the TV norm constraint is satisfied, and to achieve Joint spatial smoothness with BSC.

[0084] ③ Dual variable update: based on , , The update result updates the dual variable. The updated formula is: , Let be the dual variable after the (k+1)th iteration update.

[0085] (4) Convergence determination: Set a convergence threshold ,like Maximum number of iterations; e.g., 1000, when (Right now When the iterative change of BSC is less than the threshold or the number of iterations reaches the maximum number of iterations, the iteration stops, and the attenuation coefficient of the tissue characterized by the normalized RF signal within the sliding window is output. The estimated value And quantitative estimates of the backscattering coefficient. .

[0086] In an exemplary embodiment, S5 specifically includes: smoothing the quantitative estimate; interpolating the smoothed quantitative estimate using an interpolation method to make the dimension of the generated spatial distribution image consistent with the dimension of the ultrasound radio frequency signal, and outputting the spatial distribution image of the quantitative parameter.

[0087] In practical applications, for each quantitative parameter (such as the attenuation coefficient or backscattering coefficient), a spatial distribution map of the quantitative parameter image with a dimension lower than that of the original ultrasound radio frequency signal can be obtained. To ensure that the dimension of the quantitative parameter image remains consistent with that of the original ultrasound radio frequency signal, interpolation methods (such as bilinear or bicubic interpolation) can be used to interpolate the quantitative parameter image. The parameters corresponding to the above-mentioned parameter images are the attenuation coefficient or backscattering coefficient, which are closely related to the physical properties of the target tissue. By analyzing these parameter images, the properties of the target tissue can be analyzed.

[0088] This application offers the following advantages:

[0089] 1. High estimation accuracy: By constructing integrated data fitting terms, BSC and... The objective function of each individual L2 norm regularization constraint term and the joint TV norm space smoothing constraint term decouples the system response, propagation attenuation, and scattering signal, while scientifically determining the constraint coefficients. Compared with traditional methods, The estimation error of the is ≤8%, and the estimation error of the BSC is ≤10%, which significantly improves the accuracy of parameter estimation.

[0090] 2. Strong robustness: The constraint coefficients of the joint TV norm spatial smoothing constraint term are adaptively adjusted based on the data SNR, and the constraint coefficients of the L2 norm regularization constraint term are set based on tissue attribute matching. Moreover, the minimum value of all coefficients is greater than 0, so the constraints are effective. It can adapt to various scenarios such as low SNR and variable SNR, effectively suppress noise interference, and maintain stable estimation performance even in deep tissue imaging scenarios.

[0091] 3. Reasonable constraint settings: Separate constraints for BSC and... We set an L2 norm regularization constraint term to ensure the stability of both, and at the same time set a joint TV norm spatial smoothing constraint term to take into account the coupling relationship and joint smoothness of the two, so as to avoid extreme values ​​or local abrupt changes in the estimation results.

[0092] 4. High solution efficiency: The ADMM algorithm is used to decompose and optimize variables, and the subproblems are solved iteratively in an alternating manner. The iteration convergence speed is fast (usually convergence can be achieved in ≤500 iterations), and the computational complexity is low, which can meet the needs of real-time imaging and detection.

[0093] 5. High versatility: No need to rely on a reference phantom; only a single set of RF signals is required to complete BSC and... The joint estimation is adapted to medical ultrasound scenarios and supports parameter estimation for different probe frequencies and different human tissues. The estimation accuracy can be verified by referring to a phantom, thus improving the reliability of the method.

[0094] 6. High practicality: No additional hardware modifications are required to the existing ultrasound imaging system. The quantitative parameter estimation method for ultrasound radio frequency signals provided in this application only needs to be implemented at the software level. It has strong compatibility, and the constraint coefficient can be flexibly adjusted according to tissue properties, which facilitates its promotion and application.

[0095] This application is applied to medical ultrasound tissue characterization scenarios, specifically targeting liver tissue, to achieve BSC and... High-precision estimation, such as Figure 2 As shown, the specific steps are as follows:

[0096] Step 1. Acquire ultrasound RF signals and perform beamforming, denoising, and normalization sequentially: Using a 5MHz phased array ultrasound probe, echo RF signals from liver tissue (the tissue to be tested) and a standard ultrasound reference phantom were acquired, with a scanning depth of 20-60mm (D=512), a probe element number of N=256, and a sampling rate of 40MHz. Beamforming and 5th-order wavelet denoising were performed on the two sets of RF signals, respectively. Then, the normalized RF signal was obtained by dividing the liver tissue spectrum by the standard phantom spectrum. (512×256).

[0097] Step 2. Estimate quantitative parameters using the window sliding method: For the normalized RF signal... The analysis was performed using a sliding window method: a rectangular window with a certain side length (10 times the wavelength) was selected, and the window was zigzag-patterned across the entire target ultrasound radio frequency signal with a specific step size (1 / 4 of the window side length). For the normalized RF signal within each sliding window, the backscattering coefficient and attenuation coefficient of the tissue it represents were solved through the following sub-steps (steps 2.1-2.3).

[0098] Step 2.1. For the normalized RF signal within the sliding window, construct a joint objective function that includes data fitting and multiple regularization constraints: Construct the joint objective function Among them, the data fitting term is , The L2 norm term is The L2 norm regularization constraint term of BSC is: Joint TV Fan Items ,and , , .

[0099] Step 2.2. Determine the constraint coefficients based on the relative signal-to-noise ratio and organizational properties:

[0100] (1) Select a depth of 55-60mm (no liver tissue reflection signal, only background noise) as the noise window, and calculate the noise power at each frequency point. .

[0101] (2) Calculate the depth-frequency point And limited to the range of [0,40] dB.

[0102] (3) To Obtained by depth average .

[0103] (4) Determine the constraint coefficients: (greater than 0) ,according to Calculation of frequency-by-frequency Based on liver tissue properties, select (greater than 0) (Greater than 0), adapting to the acoustic properties of liver tissue.

[0104] Step 2.3. Iteratively solve the joint objective function using the Alternating Direction Multiplier Method (ADMM): Set , Maximum number of iterations = 1000; introduce auxiliary variables. Construct the augmented Lagrangian function; solve iteratively using alternating methods. Combined with the BSC subproblem (using the L-BFGS-B quasi-Newton method) and the u subproblem (Chambolle projection operator), update the dual variable. .

[0105] Step 2.4. After convergence, obtain the estimated values ​​of the attenuation coefficient and backscatter coefficient of the tissue characterized by the ultrasound radio frequency signal within the sliding window: After 420 iterations, the convergence condition is met, and the quantitative estimate of the attenuation coefficient of the tissue characterized by the ultrasound radio frequency signal within the sliding window is output. And quantitative estimates of the backscattering coefficient .

[0106] Step 3. Quantitative Parameter Image Generation and Post-processing: Through the window sliding operation in Step 2 above and the estimation of normalized RF signal quantitative parameters within the sliding window, a spatial distribution map of the quantitative parameter image with a dimension lower than that of the original ultrasound radio frequency signal can be obtained for each quantitative parameter (such as attenuation coefficient or backscattering coefficient). Bilinear interpolation is used to interpolate the quantitative parameter image to obtain a quantitative parameter image with the same dimension as the original ultrasound radio frequency signal, which is then used to analyze the target tissue properties.

[0107] The specific embodiments described in this application are merely preferred examples and do not constitute a limitation thereof. Those skilled in the art can make minor adjustments to the technical solutions of this application based on actual medical scenarios (such as adjusting the denoising algorithm, empirical values ​​of constraint coefficients, ADMM parameters, etc.), as long as they do not deviate from the core technical solution and ensure that all constraint coefficients are greater than 0, they all fall within the protection scope of this application.

[0108] Furthermore, this application can be implemented by a computer program (stored in a computer-readable storage medium, with all steps completed during processor execution) or by a dedicated hardware device including an ultrasound probe, processor, etc., for use in clinical medical ultrasound testing. Its estimation accuracy can be verified by a reference phantom, which does not involve the innovative content of this application.

[0109] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments. The computer device can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device stores data to be processed. The I / O interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with an external terminal via a network connection. When the computer program is executed by the processor, it implements the above-described methods.

[0110] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0111] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0112] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0113] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by hardware related to computer program instructions. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0114] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0115] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0116] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for quantitative parameter estimation of ultrasonic radio frequency signals, characterized in that, include: The ultrasound radio frequency signal reflected from the tissue under test is acquired and preprocessed to obtain a normalized RF signal; the preprocessing includes beamforming, noise reduction and normalization. The normalized RF signal is analyzed using a sliding window approach, and a joint objective function is constructed for the normalized RF signal within each sliding window. The joint objective function integrates a data fitting term, an L2 norm regularization constraint term for the backscattering coefficient, an L2 norm regularization constraint term for the attenuation coefficient, and a joint TV norm spatial smoothing constraint term for the backscattering coefficient and the attenuation coefficient. The normalized RF signal is analyzed using a sliding window approach, and a joint objective function is constructed for the normalized RF signal within each sliding window, specifically including: Using the backscattering coefficient and attenuation coefficient of the normalized RF signal within each sliding window as optimization variables, a joint objective function is constructed. The constraint coefficients of the joint objective function are determined based on the relative signal-to-noise ratio of the normalized RF signal and the tissue properties of the tissue under test, specifically including: The constraint coefficients of the joint TV norm spatial smoothing constraint term are determined based on the relative signal-to-noise ratio of the normalized RF signal, specifically including: The region of unorganized reflection signal in the normalized RF signal is selected as the noise window, and the noise power at each frequency point is calculated. Based on the noise power, the relative signal-to-noise ratio at each depth-frequency point is calculated, and the relative signal-to-noise ratio is limited to a preset range; The relative signal-to-noise ratio within the preset range is averaged along the depth dimension to obtain the frequency-specific relative signal-to-noise ratio. The constraint coefficients of the joint TV norm space smoothing constraint term are calculated by linear mapping based on the frequency-wise relative signal-to-noise ratio. The constraint coefficients of the L2 norm regularization constraint term of the attenuation coefficient and the constraint coefficients of the L2 norm regularization constraint term of the backscattering coefficient are determined based on the tissue properties of the tissue to be tested; all constraint coefficients satisfy the physical constraint condition of being greater than 0. Based on the joint objective function and all constraint coefficients, quantitative estimates of the quantitative parameters of the tissue under test are estimated, specifically including: Based on the joint objective function and all constraint coefficients, auxiliary variables are introduced, and the optimization variables are decomposed using the alternating direction multiplier method. Sub-problems are solved iteratively in alternating directions, and convergence is determined to obtain quantitative estimates of the quantitative parameters of the tissue under test. Specifically, these estimates include: By introducing auxiliary variables corresponding to the joint TV norm space smoothing constraint term, the joint objective function is transformed into an optimization problem with equality constraints. Based on the optimization problem with equality constraints, dual variables and penalty parameters are introduced to construct an augmented Lagrangian function; Based on the augmented Lagrangian function, the joint subproblem of attenuation coefficient and backscattering coefficient and the auxiliary variable subproblem are solved sequentially, and the dual variable is updated; wherein, the joint subproblem of attenuation coefficient and backscattering coefficient is solved by the conjugate gradient method or the L-BFGS-B quasi-Newton method, and the auxiliary variable subproblem is solved by the TV norm Chambolle projection operator. A convergence threshold and a maximum number of iterations are set. Iteration stops when the iterative change of the optimization variable is less than the convergence threshold or the maximum number of iterations is reached, and the quantitative estimate of the quantitative parameters of the tissue under test is output. The sub-problem includes a joint sub-problem of attenuation coefficient and backscattering coefficient and a sub-problem of auxiliary variables. The quantitative estimate includes the quantitative estimate of attenuation coefficient and the quantitative estimate of backscattering coefficient. The quantitative parameters include backscattering coefficient and attenuation coefficient. Based on the quantitative estimate, a corresponding spatial distribution image is generated; the spatial distribution image is used to analyze the properties of the tissue to be tested.

2. The method for quantitative parameter estimation of ultrasonic radio frequency signals according to claim 1, characterized in that, The normalized RF signal is analyzed using a sliding window method, specifically including: A rectangular window with a side length that is an integer multiple of the wavelength of the normalized RF signal is selected as a sliding window. The window slides across the entire target ultrasonic radio frequency signal in a zigzag manner with a preset step size to analyze the normalized RF signal. The zigzag manner is as follows: the sliding window first slides to the end along the first direction, then moves one step along the vertical direction, and then slides back along the first direction, repeating this process until the entire target ultrasonic radio frequency signal area is covered. The entire target ultrasonic radio frequency signal includes all normalized RF signals.

3. The method for quantitative parameter estimation of ultrasonic radio frequency signals according to claim 1, characterized in that, The data fitting term is used to measure the fitting error between the normalized RF signal and the model prediction value, wherein the model prediction value includes the attenuation factor of the ultrasound at the corresponding depth. The L2 norm regularization constraint term of the attenuation coefficient is used to constrain the amplitude of the attenuation coefficient; The L2 norm regularization constraint term of the backscattering coefficient is used to constrain the amplitude of the backscattering coefficient. The joint TV norm spatial smoothness constraint term includes the joint gradient of the attenuation coefficient and the backscattering coefficient, and is used to constrain the joint spatial smoothness of the attenuation coefficient and the backscattering coefficient.

4. The method for quantitative parameter estimation of ultrasonic radio frequency signals according to claim 1, characterized in that, Based on the quantitative estimate, a corresponding spatial distribution image is generated, specifically including: The quantitative estimates are then smoothed. An interpolation method is used to interpolate the smoothed quantitative estimate so that the dimension of the generated spatial distribution image is consistent with the dimension of the ultrasound radio frequency signal, and the spatial distribution image of the quantitative parameter is output.

5. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for estimating quantitative parameters of ultrasonic radio frequency signals according to any one of claims 1-4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for estimating quantitative parameters of ultrasonic radio frequency signals as described in any one of claims 1-4.

7. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for estimating quantitative parameters of ultrasonic radio frequency signals as described in any one of claims 1-4.

Images