Ultrasound transmission ct imaging method, system, computer and storage medium

Abstract

The application provides an ultrasonic transmission CT imaging method, system, computer and storage medium, which can quickly reconstruct high-resolution and high-contrast ultrasonic transmission images. The method comprises the following steps: in a layer-by-layer scanning, an annular ultrasonic transducer array calculates a TOF matrix containing transmission times of all ultrasonic paths after rotating by a specific angle each time; after completing a layer scanning, the TOF matrices at different angles are reorganized into a three-dimensional matrix according to the spatial relationship, and missing data is completed; the region to be imaged is discretized into a two-dimensional grid, each grid corresponds to a pixel point, the transmission times and path lengths of all ultrasonic paths passing through the pixel point are counted, the sound velocity contribution values of the paths are calculated, the average value is taken as the sound velocity value of the pixel point, and a two-dimensional ultrasonic transmission image is generated; the two-dimensional image is converted into a three-dimensional image through a ray casting algorithm, a depth-based weight factor is introduced to adjust the visual impact of the voxels, and the depth perception and semi-transparent rendering effect of the image are enhanced.

Description

Technical Field

[0001] This invention relates to an ultrasound transmission CT imaging method, system, computer, and storage medium. Background Technology

[0002] Capacitive micromachined ultrasonic transducers (CMUTs), as a novel transducer technology in the field of ultrasound imaging in recent years, have gradually replaced traditional piezoelectric transducers (PZTs) in many applications. CMUTs exhibit significant advantages in high-resolution ultrasound imaging due to their wide bandwidth, high sensitivity, ease of miniaturization, and large-scale integration. Furthermore, the micromachining technology used in CMUT manufacturing allows for high-density array arrangements, making them more suitable for high-precision, omnidirectional imaging applications.

[0003] Transmission ultrasound imaging (TEM) is an imaging technique based on measuring the propagation characteristics of ultrasound waves in a medium, widely used in medical imaging and industrial inspection. In this method, ultrasound waves are emitted from a transmitting transducer, pass through the measured medium, and reach a receiving transducer. Since the propagation speed and energy attenuation of ultrasound waves depend on the physical properties of the medium (such as density and sound velocity), the sound velocity distribution within the medium can be reconstructed by analyzing the time of arrival (TOF) and attenuation characteristics, thus reflecting the internal structure of the measured object. Based on the straight-ray theory, it is assumed that ultrasound waves propagate along a straight path, neglecting the effects of refraction and scattering within the medium; this simplification effectively improves computational efficiency. By measuring the flight time and path length between the transmitting and receiving transducers, and combining this with inverse problem solving, the average propagation speed of ultrasound waves in the medium can be calculated, and the sound velocity distribution map of the measured area can be reconstructed grid by grid.

[0004] However, existing technologies have certain limitations in terms of the perspective and completeness of data acquisition, especially when imaging the internal structure of complex media, where information blind spots or data loss may occur, affecting imaging accuracy and resolution. Summary of the Invention

[0005] This invention proposes an ultrasound transmission CT imaging method, system, computer, and storage medium that can rapidly reconstruct high-resolution and high-contrast ultrasound transmission images.

[0006] Firstly, a method for ultrasound transmission CT imaging is proposed, comprising: during layer-by-layer scanning, after each rotation of the annular ultrasound transducer array by a specific angle, calculating a Time-of-Flight (TOF) matrix containing the transmission times of all ultrasound paths; after completing one layer of scanning, rearranging and combining multiple TOF matrices at different rotation angles according to spatial relationships to generate a reconstructed TOF matrix, and completing the missing data in the reconstructed TOF matrix; discretizing the region to be imaged into a two-dimensional matrix composed of several grids, each grid corresponding to a pixel; for each pixel, counting all ultrasound paths passing through the pixel, and obtaining the transmission time of each ultrasound path in the completed reconstructed TOF matrix; calculating the sound velocity contribution value of each ultrasound path at the pixel based on the transmission time of each ultrasound path and its path length passing through the pixel; taking the average of the sound velocity contribution values ​​of all ultrasound paths passing through the pixel as the sound velocity value Q(r,s) of the pixel; generating a two-dimensional ultrasound transmission image based on the sound velocity values ​​Q(r,s) of all pixels; and converting the two-dimensional ultrasound transmission image into a three-dimensional image with depth perception and semi-transparent rendering effects.

[0007] In some examples, the formula for calculating the speed of sound Q(r,s) is:

[0008]

[0009] In the formula, k represents the number of ultrasound paths passing through pixel (r,s), and V i (r,s) represents the sound velocity contribution of the i-th ultrasonic path at pixel (r,s).

[0010] In some examples, the transmission time of each ultrasound path is calculated as follows: the ultrasound signal of each ultrasound path is divided into multiple signal intervals and multiple statistical models are used to predict the flight time; the AIC value corresponding to each model is calculated using the AIC algorithm, and the model with the smallest AIC value is the optimal model; an exponential weighting coefficient is generated based on the difference between the AIC values ​​of each model and the optimal model, and a dynamic fusion strategy is constructed to calculate the weighted average of the flight times predicted by different models according to the weighting coefficient, wherein the prediction result of the model whose AIC value is closer to the optimal model has a higher weight.

[0011] In some examples, the transmission time for each ultrasound path is calculated using the following formula:

[0012]

[0013] In the formula, t TOF AIC represents the transmission time of the ultrasound signal along each path; AIC(j) represents the Akaike Information Criterion value of the j-th model, used to measure the quality of the model; AIC minThe minimum AIC value among all models identifies the best model; t j Δ represents the flight time predicted by the j-th model; while Δ l =AIC l -AIC min Then it represents the AIC gap between the l-th model and the best model; exp represents the natural exponential function.

[0014] In some examples, the sound velocity contribution of each ultrasonic path at pixel (r,s) is calculated using the following formula:

[0015]

[0016] In the formula, Δt=t w -t b , representing the time of flight t of the ultrasonic signal in a liquid medium only. w Flight time t when the object being measured is present b The difference; V0 is the speed of sound in the liquid medium, and L0 is the length of the ultrasonic signal propagating in the object being measured.

[0017] In some examples, ray casting algorithms are used to convert two-dimensional ultrasound transmission images into three-dimensional images.

[0018] In some examples, during the projection of light, a weighting factor is assigned to each voxel in the 3D data field. This weighting factor is used to adjust the cumulative calculation of the color and opacity of each voxel along the light path, so that the visual influence of nearby voxels is amplified and the visual influence of distant voxels is weakened.

[0019] In a second aspect, an ultrasound transmission CT imaging system includes: a ring-shaped ultrasound transducer array coupled to the object under test via a liquid medium; a rotation and lifting control platform for controlling the rotation angle and lifting height of the ring-shaped ultrasound transducer array to achieve multi-angle layer-by-layer scanning; and a computer configured to perform the ultrasound transmission CT imaging method according to any one of claims 1-6.

[0020] Thirdly, a computer is proposed, comprising: a processor; a memory including one or more computer program modules; wherein the one or more computer program modules are stored in the memory and configured to be executed by the processor, the one or more computer program modules including instructions for implementing the ultrasound transmission CT imaging method.

[0021] Fourthly, a method for storing non-transitory computer-readable instructions is proposed, which, when executed by a computer, enables the aforementioned ultrasound transmission CT imaging method. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of multi-angle measurement of a ring ultrasonic transducer array according to an embodiment of the present invention.

[0023] Figure 2 This is a flowchart of an ultrasound transmission CT imaging method according to an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of a four-group TOF matrix data recombination method according to an embodiment of the present invention. Detailed Implementation

[0025] The ultrasonic transmission CT imaging system is designed for fully automated acquisition of ultrasonic image data of the object under test. It includes: a computer, a data acquisition system, a ring-shaped CMUT (capacitive micromechanical ultrasonic transducer) ultrasonic transducer array, a rotary lifting control platform, and a water tank.

[0026] The computer is the core of the entire system, responsible not only for controlling the operation of other hardware components but also for the crucial task of data processing. The workstation processes large volumes of ultrasound signals in real time and performs complex image reconstruction algorithms. Furthermore, it supports a user interface, allowing users to define scanning parameters such as transmission frequency, receiver gain, rotation angle, and elevation height.

[0027] The data acquisition system connects the CMUT ultrasonic transducer array to the computer. It is responsible for converting the physical quantities (such as ultrasonic signals) of the ultrasonic transducers into electrical signals, which are then provided to the computer for subsequent analysis.

[0028] The ring-shaped design of the CMUT ultrasonic transducer array allows ultrasonic waves to penetrate the object under test from multiple directions. During each scan, each transducer sequentially acts as a transmitter, sending ultrasonic pulses, while the remaining transducers act as receivers, recording the transmitted ultrasonic signals passing through the object under test.

[0029] The rotary lifting control platform precisely controls the position changes of the CMUT array. For example... Figure 1 As shown, following a predetermined procedure, the platform rotates the CMUT array at set angular intervals to achieve 360° omnidirectional coverage, and after completing one rotation, it rises a certain distance to continue scanning the next layer. This all-in-one data acquisition mode, combined with precise position adjustments, ensures omnidirectional coverage in three-dimensional space. For example, each rotation angle is one-quarter of the center angle between the two CMUT ultrasonic transducers, acquiring four different sets of data; after measuring one layer, it rises 1mm to scan the next layer.

[0030] The ultrasonic transducer array is coupled to the organ under test via a liquid medium to ensure minimal energy loss of the ultrasound waves before they reach the organ. For example, this medium can be water. The water is stored in a tank, with the detection window located on top of the tank and used to hold the organ under test. The ultrasonic transducer array is mounted on the side or sidewall of the tank (the CMUT array rotates and rises together with the tank). The specific assembly of the ultrasonic transducer array, detection window, tank, and rotation / lifting control platform is not the focus of this invention; detailed information can be found in patent document CN118787384A.

[0031] Figure 2 The ultrasound transmission CT imaging method is demonstrated. The following section discusses... Figure 2 The method shown will be described in detail.

[0032] (1) The transmission time of the ultrasonic signal is accurately extracted using the Akaike Information Criterion (AIC algorithm). The specific steps are as follows: First, the received ultrasonic signal is divided into multiple stages. For each stage, a series of different statistical models (such as linear regression models, autoregressive models, etc.) are applied to analyze the signal. Then, the AIC algorithm is used to identify the relatively stable signal intervals before and after the flight time point. This step helps to determine the exact moment when the ultrasonic wave propagates in the medium. By evaluating the performance of these different models and selecting the best model based on their AIC values, the errors that may be caused by relying on a single model are avoided. This method enhances the robustness of the time delay estimation by weighted averaging of the results of multiple models.

[0033] For each ultrasonic path, this invention employs the AIC algorithm to process the transmitted signal to extract the accurate time of arrival (TOF). This process involves selecting a suitable statistical model, identifying stable local regions in the received signal, and thus calculating the time it takes for the ultrasonic signal to travel through the object under test, i.e., TOF. The size of the time window is determined based on the speed of sound in the liquid medium (e.g., water), where j is each data point within the selected time window, n is the total number of data points within the time window (j = 1, 2, ..., n), and f... s It is the sampling frequency of the received signal.

[0034] The AIC formula is defined as follows:

[0035]

[0036] Here, AIC(j) is the Akaike Information Criterion value for the j-th model, used to select the optimal model. It takes into account the model's complexity and goodness of fit. The variance of the dataset from 1 to j is denoted as (nj-1). The smaller the variance, the better the model fit. (nj-1) represents the remaining degrees of freedom, minus j and 1 (because estimating a parameter requires losing one degree of freedom). This represents the variance of the dataset from j+1 to n.

[0037] To find the model corresponding to the minimum AIC value, the following formula is designed:

[0038]

[0039] To further enhance the robustness and accuracy of ultrasound transmission imaging technology, especially in the face of challenges such as noise interference, incomplete data, and differences in the adaptability of different models, this invention employs a robust model fusion strategy based on a weighted average method. This method dynamically adjusts the weights of each model by introducing an exponential weighting mechanism, thereby optimizing the stability and accuracy of the reconstruction results.

[0040] Specifically, the transmission time (TOF) of the signal in each ultrasound path can be calculated using the following formula:

[0041]

[0042] In the formula, t TOF AIC represents the transmission time of the ultrasound signal along each path; AIC(j) represents the Akaike Information Criterion value of the j-th model, used to measure the quality of the model; AIC min The minimum AIC value among all models identifies the best model; t j Δ represents the flight time predicted by the j-th model; while Δ l =AIC l -AIC min Then it represents the AIC gap between the l-th model and the best model; exp represents the natural exponential function.

[0043] In this framework, the present invention uses an exponential function to adjust the weights of each model, giving higher weights to models with lower AIC values ​​(i.e., better performance). This not only emphasizes the impact of more suitable models on the final results but also ensures that the sum of all model weights is 1, maintaining the rationality of the calculation process and the consistency of physical meaning. This intelligent optimization allocation method effectively alleviates the instability and error accumulation problems caused by differences in model adaptability, thereby improving the overall performance of ultrasound transmission imaging.

[0044] (2) TOF matrix completion method

[0045] The Time-of-Flight (TOF) matrix is ​​composed of the transmission time of the ultrasound signal in the ultrasound path. During the layer-by-layer scanning process, the data acquired after each rotation of the CMUT ultrasound transducer array by a certain angle is used to calculate a TOF matrix. After one layer (i.e., one slice) is scanned, the multiple TOF matrices obtained at different rotation angles are rearranged and combined to obtain a recombined TOF matrix.

[0046] Rearrangement: This may mean adjusting the positions of elements within a single Time-of-Flight (TOF) matrix to match a predefined spatial or logical order. For example, if the sensor array is physically rotated, the corresponding TOF values ​​also need to be repositioned according to the new geometry to ensure they correctly reflect the spatial relationships at the current angle. Combination: This means merging multiple independent TOF matrices into a more complete set of matrices. By combining data from different rotational positions, the potential data insufficiency from a single viewpoint can be compensated for. The combination process may include, but is not limited to, techniques such as weighted averaging and interpolation padding, aiming to utilize all available information to construct the most accurate and comprehensive possible TOF data representation. Figure 3 This is a schematic diagram of the four TOF matrix data recombination methods.

[0047] The Singular Value Thresholding (SVT) algorithm is applied to process the rows of the reconstructed Time-of-Flight (TOF) matrix to identify and repair any missing or incomplete time-of-flight data. Specifically, this process is achieved by solving the following low-rank matrix approximation optimization problem:

[0048] minimize∥X∥ *

[0049] subject to P Ω (X)=P Ω (TOF)

[0050] Where X represents the recombined TOF matrix to be completed, ∥X∥ * Let represent the nuclear norm of matrix X, which is the sum of all singular values. Minimizing the nuclear norm is a low-rank matrix approximation method because it encourages matrix X to have a lower rank. Ω (.) is the projection operator, which means that only the element value of the matrix at the known data position Ω is retained, and the other positions are set to zero. Constraint P Ω (X)=P Ω (TOF) requires that the value of matrix X at the known position Ω must be consistent with the corresponding position of the original TOF matrix (the time-of-flight data matrix directly acquired under a single scan angle), while the value at the unknown position is filled in through the optimization process.

[0051] The iterative optimization formula is as follows:

[0052]

[0053] The shrink operator is used to adjust the matrix Y. m-1 Singular values ​​are soft-thresholded to produce a new matrix X. m That is, for Y m-1 The singular values ​​obtained from the singular value decomposition (SVD) are thresholded; τ is a threshold parameter that controls the degree of singular value shrinkage; the scalar step size sequence {δ m} m≥1 Used to adjust the update magnitude in each iteration. From the initial matrix Y... 0 Starting with 0, iterate continuously until the stopping condition is met:

[0054]

[0055] Where ∈ is a fixed tolerance value, for example, 10. –4 This is a threshold used to measure relative error. When matrix X... m The iteration process terminates when the relative error at the known position Ω is less than the preset tolerance ∈. ∥.∥ F This represents the Frobenius norm, which is the square root of the sum of the squares of the matrix elements and is used to quantify the differences between matrices.

[0056] This method effectively completes the missing data in the reconstructed TOF matrix, ensuring the integrity and accuracy of the imaging, thereby improving the quality of ultrasound imaging. This, in turn, enhances the accuracy of subsequent sound velocity distribution reconstruction and improves the quality of the final generated image.

[0057] (3) Algorithm for sound velocity tomography based on straight rays

[0058] This algorithm aims to solve the inverse problem and calculate the average sound velocity in the measured object by combining the completed Time-of-Flight (TOF) matrix with the actual length of each ultrasonic wave propagation path. Subsequently, the imaging region is divided into multiple grids, and the corresponding sound velocity value is calculated for each grid to progressively reconstruct the sound velocity distribution map of the measured object, ultimately generating a high-resolution ultrasonic transmission imaging result. This is described in detail below.

[0059] For each directed line segment between the transmitting and receiving transducers, measure the flight time t when only water is present. w and the flight time t when the object being measured is present. b .

[0060] Calculate the time difference between the two:

[0061] Δt=t w -t b

[0062]

[0063] L is the propagation path length of the ultrasonic wave in a water-only medium, V0 is the speed of sound in water (a constant), and L0 is the propagation length of the ultrasonic wave within the object being measured. i It is the average speed of sound along the propagation path of the ultrasonic wave within the object being measured.

[0064] Assuming there are M transmitting transducers and M receiving transducers, then the total number of directed line segments between the transmitting and receiving transducers is M. 2 The average sound velocity V along the propagation path of the ultrasonic wave within the object being measured. i (i = 1, 2, ..., M) 2 The average sound velocity V along the propagation path of the ultrasonic wave within the object being measured can be calculated. i :

[0065]

[0066] The area to be imaged is discretized into N×N square grids, forming a two-dimensional matrix, where each element represents a pixel, denoted by coordinates (r, s) (r = 1, 2, ..., N, s = 1, 2, ..., N). The initial sound velocity value of each pixel is set to 0.

[0067] For each pixel, all ultrasonic paths passing through that pixel are counted, and the pixel's sound velocity value is updated based on the average sound velocity along these paths. Specifically, if k paths pass through pixel (r,s), then the sound velocity value Q(r,s) at that point can be represented as the weighted average or simple average of the sound velocities along these paths, i.e.:

[0068]

[0069] In the formula V i (r,s) represents the sound speed contribution of the i-th path at pixel (r,s).

[0070] Finally, based on the sound velocity value Q(r,s) of each pixel, a sound velocity distribution map of the entire imaging area can be generated, which is the ultrasonic transmission imaging result of the object under test.

[0071] (4) Achieving three-dimensional imaging through ray casting algorithm

[0072] To convert the two-dimensional ultrasound transmission images obtained in the preceding steps into three-dimensional images with depth perception and semi-transparency, this invention employs a ray projection algorithm. This process begins with a three-dimensional data field composed of multiple two-dimensional slices, obtained by scanning the object under test layer by layer using the aforementioned rotational lifting control platform. After each rotation, the CMUT array rises a certain distance (e.g., 1 mm) to continue scanning the next layer. This series of operations generates a series of continuous two-dimensional images, the so-called "slices." All these slices together constitute a three-dimensional matrix data describing the entire imaging area.

[0073] The ray casting algorithm assumes that each data point in the 3D data field has chromaticity and opacity attributes. Using parallel projection, a ray is emitted from each pixel on the projection plane, and this ray travels through the 3D dataset along a predetermined direction vector. During equidistant sampling along the ray path, the color and opacity values ​​of each sampling point are calculated using trilinear interpolation, and these values ​​are accumulated from front to back until the ray is completely absorbed or passes through the data field, thus determining the final color value of the pixel.

[0074] Specifically, for each ray, its mathematical expression can be represented as:

[0075] r(t) = r0 + td, t∈[t min ,t max ]

[0076] Where r0 represents the origin of the ray, typically corresponding to the pixel position on the projection plane; d is the ray direction vector; and t is a parameter along the ray direction. min and t max These represent the parameter values ​​for light entering and leaving the data field, respectively.

[0077] Calculate the final color value C according to the following rules. final :

[0078]

[0079] when When the cumulative calculation stops, C becomes the final color value of that pixel. i ) represents the color value of the i-th sampling point; α(t) i ) represents the opacity of the i-th sampling point; Let be the transmittance at the i-th sampling point, and represent the transmittance of light up to that point, where (1-α(t) j )) represents the transparency of that point; N in the ray casting algorithm represents the total number of points sampled along the ray direction, that is, the number of steps the ray takes to sample in the three-dimensional data field.

[0080] Furthermore, during the accumulation of color and opacity, this invention introduces a depth-based weighting factor ω to enhance the image's sense of depth and dimension. This technique is achieved through the following formula:

[0081] C(t i )←C(t i )·(1+ω·t i )

[0082] Where: t i It is between t min and t max A value between t and t represents the position parameter at a point along this ray path. i This corresponds to the position of a sampling point or voxel on the light path. The core idea of ​​this technique is to enhance the stereoscopic effect and depth perception of the image by assigning greater weight to voxels closer to the observer and less weight to voxels farther away from the observer, thereby better revealing the details of the internal structure of the object being measured.

[0083] Through the aforementioned ray projection algorithm, this invention can effectively utilize the ray projection algorithm to visualize the three-dimensional data field, thereby generating three-dimensional ultrasound transmission imaging results with depth perception and semi-transparent effects, providing intuitive and high-resolution image support for medical diagnosis and other fields.

[0084] The present invention also provides an embodiment of a computer. The computer includes a processor and a memory. The memory is used to store non-transitory computer-readable instructions (e.g., one or more computer program modules). The processor is used to execute the non-transitory computer-readable instructions, which, when executed by the processor, can perform one or more steps in the ultrasound transmission CT imaging method described above. The memory and the processor can be interconnected via a bus system and / or other forms of connection mechanisms.

[0085] For example, a processor can be a central processing unit (CPU), a graphics processing unit (GPU), or other form of processing unit with data processing and / or program execution capabilities. For instance, a CPU can be based on x86 or ARM architectures. A processor can be a general-purpose processor or a special-purpose processor, and it can control other components in a computer to perform desired functions.

[0086] For example, memory can include any combination of one or more computer program products, which can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory can include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory can include, for example, read-only memory (ROM), hard disk, erasable programmable read-only memory (EPROM), compact optical disc read-only memory (CD-ROM), USB storage, flash memory, etc. One or more computer program modules can be stored on the computer-readable storage medium, and the processor can run one or more computer program modules to implement various functions of the computer.

[0087] This invention also provides a computer-readable storage medium for storing non-transitory computer-readable instructions that, when executed by a computer, can implement one or more steps in the aforementioned ultrasound transmission CT imaging method. When the ultrasound transmission CT imaging method provided in this embodiment is implemented in software and sold or used as a standalone product, it can be stored in a computer-readable storage medium. For further details regarding the storage medium, please refer to the corresponding description of memory in the computer system above; it will not be repeated here.

Claims

1. A method for ultrasound transmission CT imaging, characterized in that, include: During the layer-by-layer scanning process, after each rotation of the annular ultrasonic transducer array by a specific angle, a Time-of-Flight (TOF) matrix is ​​calculated, which contains the transmission time of all ultrasonic paths. After completing one layer of scanning, the multiple TOF matrices under different rotation angles are rearranged and combined according to spatial relationships to generate a recombined TOF matrix, and the missing data in the recombined TOF matrix is ​​filled in. The region to be imaged is discretized into a two-dimensional matrix composed of several grids, with each grid corresponding to a pixel. For each pixel, all ultrasound paths passing through the pixel are counted, and the transmission time of each ultrasound path in the completed reconstructed TOF matrix is ​​obtained. Based on the transmission time of each ultrasound path and its path length passing through the pixel, the sound velocity contribution value of each ultrasound path at the pixel is calculated. The average sound velocity contribution value of all ultrasound paths passing through the pixel is taken as the sound velocity value Q(r,s) of the pixel. A two-dimensional ultrasound transmission image is generated based on the sound velocity values ​​Q(r,s) of all pixels. Convert two-dimensional ultrasound transmission images into three-dimensional images with depth perception and semi-transparent rendering effects.

2. The ultrasonic transmission CT imaging method according to claim 1, characterized in that, The formula for calculating the speed of sound Q(r,s) is as follows: In the formula, k represents the number of ultrasound paths passing through pixel (r,s), and V i (r,s) represents the sound velocity contribution of the i-th ultrasonic path at pixel (r,s).

3. The ultrasonic transmission CT imaging method according to claim 1, characterized in that, The transmission time of each ultrasound path is calculated as follows: the ultrasound signal of each ultrasound path is divided into multiple signal intervals and multiple statistical models are used to predict the flight time; the AIC value of each model is calculated using the AIC algorithm, and the model with the smallest AIC value is the optimal model; based on the difference between the AIC values ​​of each model and the optimal model, an exponential weighting coefficient is generated, and a dynamic fusion strategy is constructed. The flight times predicted by different models are weighted and averaged according to the weighting coefficient, where the prediction result of the model whose AIC value is closer to the optimal model has a higher weight.

4. The ultrasound transmission CT imaging method according to claim 3, characterized in that, The transmission time of each ultrasound path is calculated using the following formula: In the formula, t TOF AIC represents the transmission time of the ultrasound signal along each path; AIC(j) represents the Akaike Information Criterion value of the j-th model, used to measure the quality of the model; AIC min The minimum AIC value among all models identifies the best model; t j Δ represents the flight time predicted by the j-th model; while Δ l =AIC l -AIC min Then it represents the AIC gap between the l-th model and the best model; exp represents the natural exponential function.

5. The ultrasonic transmission CT imaging method according to claim 1, characterized in that, The formula for calculating the sound velocity contribution of each ultrasonic path at pixel (r,s) is as follows: In the formula, Δt=t w -t b , representing the time of flight t of the ultrasonic signal in a liquid medium only. w Flight time t when the object being measured is present b The difference; V0 is the speed of sound in the liquid medium, and L0 is the length of the ultrasonic signal propagating in the object being measured.

6. The ultrasonic transmission CT imaging method according to claim 1, characterized in that, A ray projection algorithm is used to convert two-dimensional ultrasound transmission images into three-dimensional images.

7. The ultrasound transmission CT imaging method according to claim 6, characterized in that, During the projection of light, a weighting factor is assigned to each voxel in the three-dimensional data field. This weighting factor is used to adjust the cumulative calculation of the color and opacity of each voxel along the light path, so that the visual influence of nearby voxels is amplified and the visual influence of distant voxels is weakened.

8. An ultrasonic transmission CT imaging system, characterized in that, Include: A ring-shaped ultrasonic transducer array, which is coupled to the object under test through a liquid medium; A rotating and lifting control platform is used to control the rotation angle and lifting height of the annular ultrasonic transducer array to achieve multi-angle layer-by-layer scanning; and A computer configured to perform the ultrasound transmission CT imaging method according to any one of claims 1-6.

9. A computer, characterized in that, Include: processor; Memory, including one or more computer program modules; The one or more computer program modules are stored in the memory and configured to be executed by the processor, and the one or more computer program modules include instructions for implementing the ultrasound transmission CT imaging method according to any one of claims 1-6.

10. A method for storing non-transitory computer-readable instructions, characterized in that, When the non-transitory computer-readable instructions are executed by a computer, the ultrasound transmission CT imaging method according to any one of claims 1-6 can be implemented.

Images