Microbubble image generation method and device, model training method and device and super-resolution ultrasonic imaging method and device

Abstract

The invention relates to the technical field of ultrasonic imaging, and provides a microbubble image generation method and device, a model training method and device and a super-resolution ultrasonic imaging method and device, and the ultrasonic imaging method comprises the steps: obtaining multiple frames of first ultrasonic images of a target object; inputting the multiple frames of first ultrasonic images into a super-resolution ultrasonic imaging model to generate a super-resolution ultrasonic image of the target object; the super-resolution ultrasonic imaging model is obtained by training a deep neural network by using a microbubble image data set, the microbubble image data set comprises simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on the microbubble images, and the microbubble images are obtained by converting corresponding second ultrasonic images; the second ultrasonic image is obtained by filtering a background signal in a corresponding third ultrasonic image, and the third ultrasonic image is a microbubble image of the biological tissue injected with the microbubble contrast agent under different imaging parameter conditions. According to the embodiment of the invention, the imaging precision of super-resolution ultrasonic imaging can be improved.

Description

Technical Field

[0001] This specification relates to the technical field of ultrasonic imaging, and in particular, to a method and device for generating microbubble images, training models, and super-resolution ultrasonic imaging. Background Art

[0002] Medical ultrasonic imaging is an important clinical auxiliary diagnosis technology. In particular, color Doppler ultrasonic imaging technology is an important auxiliary means for detecting cardiovascular diseases in the clinic. However, traditional ultrasonic imaging is restricted by the limitations of imaging depth and imaging resolution, and it is impossible to achieve in-vivo deep imaging while ensuring imaging resolution. In recent years, the proposal of super-resolution ultrasonic imaging technology has solved the above problems. This technology can improve the imaging resolution by nearly ten times while ensuring large-depth imaging.

[0003] The single-frame imaging time of super-resolution ultrasonic imaging often takes several minutes, so its clinical application is limited. The imaging method based on deep neural network can effectively solve the imaging speed problem. However, deep neural network is a data-driven method and requires high-quality simulated microbubble images to achieve high-quality network training. Currently, most of the simulation methods of deep learning are based on ideal conditions, with a large deviation from the actual situation, which affects the prediction accuracy of the trained deep neural network, and further affects the imaging accuracy of super-resolution ultrasonic imaging. Summary of the Invention

[0004] The purpose of the embodiments of this specification is to provide a method and device for generating microbubble images, training models, and super-resolution ultrasonic imaging, so as to obtain high-quality simulated microbubble images and improve the imaging accuracy of super-resolution ultrasonic imaging.

[0005] To achieve the above object, on the one hand, the embodiments of this specification provide a method for generating microbubble images, including:

[0006] Obtaining multiple frames of first ultrasonic images of a target object;

[0007] Inputting the multiple frames of first ultrasonic images into a super-resolution ultrasonic imaging model to generate a super-resolution ultrasonic image of the target object; the super-resolution ultrasonic imaging model is obtained by training a deep neural network using a microbubble image dataset. The microbubble image dataset includes simulated microbubble images generated by a microbubble image generator. The microbubble image generator is obtained by training a generative network based on microbubble images. The microbubble images are obtained by converting corresponding second ultrasonic images, and the second ultrasonic images are obtained by filtering background signals from corresponding third ultrasonic images. The third ultrasonic images are microbubble images of biological tissues injected with microbubble contrast agents under different imaging parameter conditions.

[0008] In the super-resolution ultrasound imaging method according to the embodiments of the present specification, the imaging parameter conditions include imaging frequency, the number of ultrasonic transducer array elements, the width of the ultrasonic transducer array elements, and / or the imaging pulse length.

[0009] In the super-resolution ultrasound imaging method according to the embodiments of the present specification, filtering the background signal in the corresponding third ultrasound image includes:

[0010] Filtering the third ultrasound image to filter out the biological tissue signal and the noise signal in the third ultrasound image.

[0011] In the super-resolution ultrasound imaging method according to the embodiments of the present specification, the microbubble image is obtained by the following method:

[0012] Identifying local maximum pixel points or centroids in the third ultrasound image; the gray value of the local maximum pixel points or centroids is not less than the gray values of the surrounding pixel points;

[0013] Taking out each local maximum pixel point or centroid in the third ultrasound image and its surrounding pixel points to form a second ultrasound image;

[0014] Performing a correlation operation on the second ultrasound image and the point spread function to obtain a correlation factor between the second ultrasound image and the point spread function;

[0015] Judging whether the correlation factor is greater than a correlation factor threshold;

[0016] If the correlation factor is not greater than the correlation factor threshold, screening out the second ultrasound image;

[0017] If the correlation factor is greater than the correlation factor threshold, calculating the central position of each local maximum pixel point or centroid in the second ultrasound image by a microbubble positioning method;

[0018] Translating each local maximum pixel point or centroid in the second ultrasound image to the corresponding central position to obtain a microbubble image.

[0019] In the super-resolution ultrasound imaging method according to the embodiments of the present specification, performing a correlation operation on the second ultrasound image and the point spread function includes:

[0020] According to the formula Performing a correlation operation on the second ultrasound image and the point spread function;

[0021] where c is the correlation factor, N is the number of local maximum pixel points or centroids in the second ultrasound image, S(i) is the gray value of the i-th local maximum pixel point or centroid in the second ultrasound image, G(i) is the gray value of the i-th pixel point in the point spread function, μ Sis the average gray value of the local maximum pixel points or centroids in the second ultrasonic image, and σ S is the standard deviation of the gray values of the local maximum pixel points or centroids in the second ultrasonic image, and μ G is the average gray value of the pixel points in the point spread function, and σ G is the standard deviation of the gray values of the pixel points in the point spread function, and * represents convolution.

[0022] In the super-resolution ultrasonic imaging method according to the embodiments of the present specification, the point spread function includes:

[0023]

[0024] wherein, G is the point spread function, e is the natural constant, x and y are the horizontal and vertical coordinates of the pixel points in the point spread function respectively, and σ x is the horizontal width of G, and d is the imaging depth, s is the number of ultrasonic transducer array elements, p is the width of the ultrasonic transducer array elements, and σ y is the vertical width of G, and σ y = l, and l is the length of the imaging pulse.

[0025] In the super-resolution ultrasonic imaging method according to the embodiments of the present specification, taking out each local maximum pixel point or centroid and its surrounding pixel points in the third ultrasonic image includes:

[0026] Identifying whether the surrounding pixel points of each local maximum pixel point or centroid in the third ultrasonic image contain local maximum pixel points or centroids;

[0027] If it contains local maximum pixel points or centroids, when taking out each local maximum pixel point or centroid and its surrounding pixel points in the third ultrasonic image, retain the local maximum pixel points or centroids among the surrounding pixel points.

[0028] In the super-resolution ultrasonic imaging method according to the embodiments of the present specification, the microbubble image generator generates a simulated microbubble image in the following manner:

[0029] Obtain imaging parameter conditions;

[0030] Call the microbubble image generator according to the imaging parameter conditions to generate a corresponding simulated microbubble image;

[0031] Convolve the simulated microbubble image with the scatter plot to obtain a simulated microbubble image with a random distribution;

[0032] Assign microbubble velocities to the simulated microbubble image with a random distribution to obtain a simulated microbubble image in motion under the imaging parameter conditions;

[0033] Repeat the above steps of generating simulated microbubble images to obtain images of simulated microbubbles in motion under different imaging parameter conditions.

[0034] On the other hand, an embodiment of this specification also provides a method for generating microbubble images, including:

[0035] Collect third ultrasonic images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0036] Filter out background signals in the third ultrasonic image to obtain a second ultrasonic image;

[0037] Convert the second ultrasonic image into a microbubble image;

[0038] Train a generative network based on the microbubble image to obtain a microbubble image generator;

[0039] Generate simulated microbubble images according to the microbubble image generator.

[0040] On the other hand, an embodiment of this specification also provides a method for training a model of a super-resolution ultrasonic imaging model, including:

[0041] Obtain a microbubble image data set; the microbubble image data set includes simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on microbubble images, the microbubble images are converted from corresponding second ultrasonic images, the second ultrasonic images are obtained by filtering out background signals in corresponding third ultrasonic images, and the third ultrasonic images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0042] Use the microbubble image data set to train a deep neural network to obtain a super-resolution ultrasonic imaging model.

[0043] On the other hand, an embodiment of this specification also provides a super-resolution ultrasonic imaging device, including:

[0044] A first acquisition module for acquiring multiple frames of first ultrasonic images of a target object;

[0045] A first generation module, configured to input the multiple frames of first ultrasound images into a super-resolution ultrasound imaging model to generate a super-resolution ultrasound image of the target object; the super-resolution ultrasound imaging model is obtained by training a deep neural network using a microbubble image dataset, the microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on microbubble images, the microbubble images are converted from corresponding second ultrasound images, and the second ultrasound images are obtained by filtering background signals from corresponding third ultrasound images, where the third ultrasound images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions.

[0046] On the other hand, an embodiment of this specification further provides a microbubble image generation device, including:

[0047] An acquisition module, configured to acquire third ultrasound images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0048] A filtering module, configured to filter background signals from the third ultrasound images to obtain second ultrasound images;

[0049] A conversion module, configured to convert the second ultrasound images into microbubble images;

[0050] A first training module, configured to train a generative network based on the microbubble images to obtain a microbubble image generator;

[0051] A second generation module, configured to generate simulated microbubble images according to the microbubble image generator.

[0052] On the other hand, an embodiment of this specification further provides a model training device for a super-resolution ultrasound imaging model, including:

[0053] A second acquisition module, configured to acquire a microbubble image dataset; the microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on microbubble images, the microbubble images are converted from corresponding second ultrasound images, and the second ultrasound images are obtained by filtering background signals from corresponding third ultrasound images, where the third ultrasound images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0054] A second training module, configured to train a deep neural network using the microbubble image dataset to obtain a super-resolution ultrasound imaging model.

[0055] On the other hand, an embodiment of this specification further provides a computer device, including a memory, a processor, and a computer program stored on the memory. When the computer program is run by the processor, it executes the instructions of the above method.

[0056] On the other hand, an embodiment of this specification also provides a computer storage medium, on which a computer program is stored. When the computer program is run by a processor of a computer device, it executes the instructions of the above method.

[0057] On the other hand, an embodiment of this specification also provides a computer program product, which includes a computer program. When the computer program is run by a processor of a computer device, it executes the instructions of the above method.

[0058] As can be seen from the technical solutions provided by the embodiments of this specification above, in the embodiments of this specification, under different imaging parameter conditions, after the ultrasonic image is filtered by the background signal, a cleaner ultrasonic image can be obtained. Then, these cleaner ultrasonic images are converted into microbubble images, and the generated microbubble images are used as inputs to train a generative network to obtain a microbubble image generator. Then, a simulated microbubble image is generated according to the microbubble image generator. Since the input during training is the microbubble image converted from the clean ultrasonic image, and the initial training model is a generative network, the simulated microbubble image generated according to the microbubble image generator has a high fidelity and is close to the real microbubble image, that is, a high-quality simulated microbubble image is obtained. In this way, a high-quality data set is provided for subsequent training of the super-resolution ultrasonic imaging model, which is conducive to training a super-resolution ultrasonic imaging model with higher prediction accuracy, and ultimately conducive to improving the super-resolution ultrasonic imaging accuracy. Description of the Drawings

[0059] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings:

[0060] Figure 1 Shows a flowchart of a microbubble image generation method in some embodiments of this specification;

[0061] Figure 2 Shows Figure 1 A flowchart of converting a microbubble image from a corresponding second ultrasonic image in the method shown;

[0062] Figure 3 Shows a schematic diagram of local maximum pixel points or centroids and their surrounding pixel points in an exemplary embodiment of this specification;

[0063] Figure 4 Shows Figure 1Flowchart of generating a simulated microbubble image according to the microbubble image generator in the method shown;

[0064] Figure 5 Shows Figure 1 Schematic diagram of generating a simulated microbubble image by the microbubble image generator in the method shown;

[0065] Figure 6 Flowchart of the model training method of the super-resolution ultrasound imaging model in some embodiments of the present specification;

[0066] Figure 7 Schematic diagram of the model training of the super-resolution ultrasound imaging model in an exemplary embodiment of the present specification;

[0067] Figure 8 Flowchart of the super-resolution ultrasound imaging method in some embodiments of the present specification;

[0068] Figure 9 Schematic diagram of super-resolution ultrasound imaging in an exemplary embodiment of the present specification;

[0069] Figure 10 Structure block diagram of the super-resolution ultrasound imaging device in some embodiments of the present specification;

[0070] Figure 11 Structure block diagram of the microbubble image generating device in some embodiments of the present specification;

[0071] Figure 12 Structure block diagram of the model training device of the super-resolution ultrasound imaging model in some embodiments of the present specification;

[0072] Figure 13 Structure block diagram of the computer device in some embodiments of the present specification.

[0073]

Explanation of the reference numerals

[0074] 11. First acquisition module;

[0075] 12. First generation module;

[0076] 21. Acquisition module;

[0077] 22. Filtering module;

[0078] 23. Conversion module;

[0079] 24. First training module;

[0080] 25. Second generation module;

[0081] 31. Second acquisition module;

[0082] 32. Second training module;

[0083] 1302. Computer device;

[0084] 1304. Processor;

[0085] 1306. Memory;

[0086] 1308. Driving mechanism;

[0087] 1310. Input / output interface;

[0088] 1312. Input device;

[0089] 1314. Output device;

[0090] 1316. Rendering device;

[0091] 1318. Graphical user interface;

[0092] 1320. Network interface;

[0093] 1322. Communication link;

[0094] 1324. Communication bus. Detailed implementation

[0095] In order to enable those skilled in the art of this technology to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the scope of protection of this specification.

[0096] It should be noted that in the embodiments of this specification, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved are all information and data that have been authorized and consented to by the user and fully authorized by all parties, that is, the acquisition, transmission, storage, use, processing, etc. of data in the technical solutions of this application all comply with the relevant provisions of national laws and regulations.

[0097] It should be noted that in the embodiments of this application, some industry-existing solutions such as certain software, components, models, etc. may be mentioned. They should be regarded as exemplary, and their purpose is only to illustrate the feasibility in the implementation of the technical solutions of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0098] Limited by the physical diffraction limit, it is difficult for the resolution of ultrasonic imaging to exceed half of the wavelength of the imaging sound wave. Although the imaging resolution can be improved by reducing the wavelength, smaller wavelengths will cause stronger energy attenuation of the sound wave by tissues, which limits the imaging depth of ultrasonic imaging. Therefore, traditional ultrasonic imaging cannot achieve super-resolution imaging (such as micro-blood flow imaging in vivo), and the emergence of super-resolution ultrasound has broken this restriction. Taking micro-blood flow imaging as an example, super-resolution ultrasound imaging locates and tracks microbubbles of ultrasound contrast agents diluted in blood vessels at the sub-wavelength level, and indirectly reflects the blood flow movement through the movement trajectories of microbubbles with a particle size of only 1-2 microns, realizing vascular imaging beyond the physical resolution.

[0099] However, traditional centroid-based imaging methods can only locate sparse microbubble signals, so the number of microbubbles that can be located in a single-frame image is limited. Often, super-resolution ultrasound imaging requires collecting tens of thousands of frames of data for several minutes for microbubble localization and tracking to achieve one imaging, which greatly limits the application of super-resolution ultrasound imaging. In recent years, methods based on deep neural networks have been applied to super-resolution ultrasound imaging, and these works have all proved that deep neural networks can effectively improve the imaging quality and imaging speed of super-resolution ultrasound imaging. However, deep neural networks are a data-driven method and require high-quality simulated microbubble images to achieve high-quality network training. Currently, most of the simulation methods for deep learning are based on ideal conditions, with a large deviation from the actual situation, thus affecting the prediction accuracy of the trained deep neural network, and further affecting the imaging accuracy of super-resolution ultrasound imaging.

[0100] In view of this, in order to obtain high-quality simulated microbubble images and improve the imaging accuracy of super-resolution ultrasound imaging, the embodiments of this specification provide an improved method for generating microbubble images, a method for training a super-resolution ultrasound imaging model, a super-resolution ultrasound imaging method, and a device.

[0101] Reference Figure 1 As shown, some embodiments of this specification provide a method for generating microbubble images, which may include the following steps:

[0102] Step 101, collect third ultrasonic images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions.

[0103] Step 102, filter out the background signal in the third ultrasonic image to obtain a second ultrasonic image.

[0104] Step 103, convert the second ultrasonic image into a microbubble image.

[0105] Step 104, train a generative network based on the microbubble image to obtain a microbubble image generator.

[0106] Step 105: Generate a simulated microbubble image according to the microbubble image generator.

[0107] In the embodiments of the present specification, after the ultrasound images under different imaging parameter conditions are filtered for background signals, a cleaner ultrasound image can be obtained. Then, these cleaner ultrasound images are converted into microbubble images. Using the converted microbubble images as input, a generative network is trained to obtain a microbubble image generator. Then, a simulated microbubble image is generated according to the microbubble image generator. Since the input during training is the microbubble image converted from the cleaner ultrasound image, and the initial training model is a generative network, the simulated microbubble image generated according to the microbubble image generator has a high fidelity and is close to the real microbubble image, that is, a high-quality simulated microbubble image is obtained. In this way, a high-quality data set is provided for the subsequent training of the super-resolution ultrasound imaging model, which is conducive to training a super-resolution ultrasound imaging model with higher prediction accuracy, and ultimately conducive to improving the super-resolution ultrasound imaging accuracy.

[0108] In some embodiments of the present specification, the biological tissue injected with the microbubble contrast agent can be one or more different biological tissues, such as different parts of the human body (such as the head, hand, leg, etc.).

[0109] In some embodiments of the present specification, in order to obtain as diverse third ultrasound images as possible, the third ultrasound images of the biological tissue injected with the microbubble contrast agent under different imaging parameter conditions can be collected. Among them, the third ultrasound image can be, for example, a B-mode ultrasound image (i.e., a B-ultrasound image). The imaging parameters can include, for example, but are not limited to, the imaging frequency f, the number of ultrasonic transducer array elements s, the width p of the ultrasonic transducer array elements, and / or the imaging pulse length l, etc., which are imaging parameters that have an important impact on the imaging accuracy. Among them, the ultrasonic transducer array element refers to: an ultrasonic transducer array formed by a plurality of ultrasonic transducers (such as a rectangular array or a square array); correspondingly, the width of the ultrasonic transducer array element refers to: the width of the ultrasonic transducer array. For example, taking the imaging parameters as the imaging frequency f, the number of ultrasonic transducer array elements s, the width p of the ultrasonic transducer array elements, and / or the imaging pulse length l as an example, when any one or more of the imaging frequency f, the number of ultrasonic transducer array elements s, the width p of the ultrasonic transducer array elements, and / or the imaging pulse length l change in value, a new imaging parameter condition is formed.

[0110] In some embodiments of this specification, the collected B-mode image not only contains microbubble signals, but also contains background signals (such as biological tissue signals, noise signals); among them, the tissue signal in the background signal has a strong echo and will cause great interference to the microbubble signal (it may even mask the microbubble signal); therefore, filtering (such as band-pass filtering, or singular value decomposition filtering, etc.) can be performed on the B-mode ultrasound image to filter out the biological tissue signal and noise signal in the B-mode ultrasound image as much as possible.

[0111] Specifically, although the contrast agent microbubbles effectively improve the echo intensity of blood signals, the signal of a single microbubble is still weaker than the tissue signal. Because the concentration of microbubbles in blood vessels is low, when microbubbles enter the microvessels, the microbubble signal will be masked by the tissue signal and cannot be observed in the B-mode image, that is, the Tissue Shadowing phenomenon. The noise signal will also cause further interference to the positioning and tracking of microbubbles. The echo characteristics of many speckle noises in a single B-mode image are close to the microbubble signal. If the noise signal cannot be correctly filtered, it will lead to subsequent positioning and tracking errors. Therefore, in order to achieve accurate super-resolution ultrasound imaging, it is first necessary to filter out the tissue and noise signals in the image to obtain microbubble and noise signals.

[0112] For example, in an exemplary embodiment, the passband width of the band-pass filter can be selected to be 50 Hz to 250 Hz, that is, the signal parts less than or equal to 50 Hz and greater than or equal to 250 Hz in the B-mode ultrasound image are filtered out.

[0113] For example, in another exemplary embodiment, taking singular value decomposition filtering (SVD) as an example, by using SVD filtering to filter the B-mode ultrasound image, efficient microbubble filtering can be achieved in both low-frequency and high-frequency imaging cases.

[0114] Among them, SVD filtering utilizes the amplitude difference between different signals. By decomposing and screening the components of the spatio-temporal signal, the decomposition of tissue, blood and noise signals is realized, and the components can be reconstructed to obtain the images of the components. As described before, there are differences in amplitude and coherence among tissue signals, blood flow signals and noise signals. Therefore, the principle of the commonly used singular value decomposition filter currently is to utilize the energy difference between different components, and select the signal component with medium energy as the blood flow signal by screening the singular values. The tissue signal has a large amplitude and strong coherence, so the image reconstructed by the larger singular values and singular value vectors is the tissue signal. And the smaller singular values and singular value vectors often represent noise signals with smaller amplitudes and weaker coherence. So select an intermediate subset [N1, N2] (1 < N1 < N2 ≤ N t)A low-amplitude and low-spatial coherence blood and microbubble signal can be obtained, and then the inverse reconstruction of singular value decomposition is performed on all the filtered blood and microbubble signal components to obtain microbubble and blood images. For microbubbles, in an exemplary embodiment, filtering out the first 10% of the largest singular values can effectively improve the distinguishability of microbubbles. In some embodiments of the present specification, the microbubble images converted from the corresponding second ultrasound images can be used for the model training of the generative network to train a microbubble image generator. Among them, the second ultrasound image refers to the ultrasound image obtained after band-pass filtering processing.

[0115] Reference Figure 2 As shown, in some embodiments of the present specification, converting the third ultrasound image into a microbubble image may include the following steps:

[0116] Step 201, identify the local maximum pixel points in the third ultrasound image; the gray value of the local maximum pixel points is not less than the gray values of its surrounding pixel points.

[0117] As Figure 3 shown, since the gray value of pixel point 5 is not less than the gray values of its surrounding pixel points (pixel points 1, 2, 3, 4, 6, 7, 8, 9), pixel point 5 can be used as a local maximum pixel point in the third ultrasound image.

[0118] Step 202, take out each local maximum pixel point in the third ultrasound image and its surrounding pixel points to form a second ultrasound image.

[0119] The second ultrasound image refers to the ultrasound image formed after taking out each local maximum pixel point in the third ultrasound image and its surrounding pixel points.

[0120] In some embodiments of the present specification, a small image area of 4×4λ (λ is the wavelength of the ultrasound imaging signal) is taken out around each local maximum pixel point. Since the image size of microbubbles is generally a cluster with a width of 1-2λ, taking out the part of 4×4λ around the local maximum pixel point can better retain the image features of microbubbles.

[0121] In some other embodiments of the present specification, each local maximum pixel point in the third ultrasound image and its surrounding pixel points can also be taken out in units of pixel points.

[0122] For example, taking Figure 3 the local maximum pixel point shown as an example, since pixel point 5 is the local maximum pixel point, pixel point 5 and its surrounding pixel points (pixel points 1, 2, 3, 4, 6, 7, 8, 9) can be taken out together.

[0123] In some other embodiments of this specification, taking out each local maximum pixel point and its surrounding pixel points in the third ultrasonic image may further include:

[0124] Identifying whether the surrounding pixel points of each local maximum pixel point in the third ultrasonic image contain local maximum pixel points; if they contain local maximum pixel points, when taking out each local maximum pixel point and its surrounding pixel points in the third ultrasonic image, retaining the local maximum pixel points among the surrounding pixel points; otherwise, there is no need to retain them. In this way, misdeleting local maximum pixel points in the third ultrasonic image can be avoided.

[0125] For example, taking Figure 3 the shown local maximum pixel point as an example, pixel point 5 is a local maximum pixel point. If the gray value of pixel point 9 is the same as or equivalent to the gray value of pixel point 5, and the gray value of pixel point 9 is also not less than the gray values of its surrounding pixel points, then when taking out pixel point 5 and its surrounding pixel points, pixel point 9 is retained (that is, pixel point 9 will not be taken out together).

[0126] Step 203: Perform a correlation operation on the second ultrasonic image and the point spread function to obtain the correlation factor between the second ultrasonic image and the point spread function.

[0127] Among them, the point spread function G can be an ideal Gaussian function. Taking the central pixel as the origin, the coordinates of other pixel points are given as (x, y), x, y ∈ (-2λ, 2λ), and according to the imaging parameters, the lateral width σ x and the longitudinal width σ y of this Gaussian function can be given; among them, d is the imaging depth, s is the number of ultrasonic transducer array elements, p is the width of the ultrasonic transducer array element, σ y = l, where l is the length of the imaging pulse. Therefore, the point spread function G can be expressed as: Among them, e is the natural constant, and x and y are the horizontal and vertical coordinates of the pixel points in the point spread function respectively.

[0128] In some embodiments of this specification, performing a correlation operation on the second ultrasonic image and the point spread function may include: According to the formula performing a correlation operation on the second ultrasonic image and the point spread function; where c is the correlation factor, N is the number of local maximum pixel points in the second ultrasonic image, S(i) is the gray value of the i-th local maximum pixel point in the second ultrasonic image, G(i) is the gray value of the i-th pixel point in the point spread function, μ S is the average gray value of the local maximum pixel points in the second ultrasonic image, σ S is the standard deviation of the gray values of the local maximum pixel points in the second ultrasonic image, μG is the mean gray value of the pixel points in the point spread function, and σ G is the standard deviation of the gray value of the pixel points in the point spread function, and * represents convolution.

[0129] Step 204, determine whether the correlation factor is greater than the correlation factor threshold. If it is greater than the correlation factor threshold, execute Step 205; otherwise, execute Step 207.

[0130] Step 205, calculate the central position of each local maximum pixel point in the second ultrasonic image through the microbubble positioning method.

[0131] If the correlation factor is greater than the correlation factor threshold, it can be considered that the second ultrasonic image conforms to the point spread law (gray value distribution law) of the point spread function and belongs to a high-quality image. Therefore, the second ultrasonic image can be retained, and the central position (sub-micron level central position) of each local maximum pixel point in the second ultrasonic image can be calculated through the microbubble positioning method. In some embodiments of this specification, the microbubble positioning method can adopt, for example, the Gaussian fitting method, etc.

[0132] Step 206, translate each local maximum pixel point in the second ultrasonic image to the corresponding central position to obtain a microbubble image.

[0133] Step 207, screen out the second ultrasonic image.

[0134] If the correlation factor is greater than the correlation factor threshold, it can be considered that the second ultrasonic image conforms to the point spread law (gray value distribution law) of the point spread function and does not belong to a high-quality image. Therefore, this image can be discarded.

[0135] In some other embodiments of this specification, when converting the third ultrasonic image into a microbubble image, it is also possible to Figure 2 replace the search for local maximum pixel points in the shown embodiments with the search for the centroid; correspondingly, for the search for the centroid, the centroid positioning algorithm, etc. can be adopted to implement. Among them, the centroid refers to the geometric center of a polygon, and the average value of the vertex coordinates of the polygon is the coordinate of the centroid. In the embodiments of this specification, the average value of the gray values of the vertices of the polygon is the gray value of the centroid. The general principle of the centroid positioning algorithm is as follows: The beacon node periodically broadcasts beacon packets to neighboring nodes. The beacon packets contain the identification number and location information of the beacon node. When the unknown node receives the beacon packets from different beacon nodes and the number exceeds a certain threshold k or after receiving for a certain period of time, it determines its own position as the centroid of the polygon formed by these beacon nodes. As described above, Figure 2 shows the processing logic for converting a single third ultrasonic image into a microbubble image; and for each third ultrasonic image, by adopting this conversion method, the corresponding microbubble image can be converted.

[0136] A generative network is a special type of neural network. Its principle is to enable a deep neural network to learn training data images and generate images similar to the training data. In some embodiments of this specification, the generative network may include a generative adversarial network (GAN), a diffusion model, and the like. In this specification, the input data of the network is a single microbubble signal. At the same time, because there are significant differences in microbubble images under different imaging parameter conditions (y = [f, s, p, l]), the ultrasonic imaging parameters are also used as a conditional vector and taken as another input of the network in this specification, so that the network can generate microbubble images under different parameters.

[0137] For example, in some embodiments of this specification, taking the generative adversarial network as an example, the generative adversarial network consists of two networks (a generator and a discriminator). The generator can generate fake images (i.e., fake microbubble images) similar to the training data (i.e., real microbubble images), while the discriminator is responsible for identifying whether the images generated by the generator are real images; the generator and the discriminator are alternately trained in the continuous generation and recognition confrontation until the discriminator has difficulty identifying whether the images generated by the generator are real images. At this time, the obtained generator is the microbubble image generator.

[0138] Reference Figure 4 As shown, in some embodiments of this specification, generating a simulated microbubble image according to the microbubble image generator may include the following steps:

[0139] Step 401, obtain the imaging parameter conditions.

[0140] In some embodiments of this specification, obtaining the imaging parameter conditions may be to generate the imaging parameter conditions in a manner of randomly taking values according to the parameters or according to the parameter variation law (such as stepping from the minimum value to the maximum value, etc.).

[0141] Step 402, call the microbubble image generator according to the imaging parameter conditions to generate a corresponding simulated microbubble image.

[0142] By inputting the imaging parameter conditions into the microbubble image generator, the microbubble image generator can use this as a constraint condition to generate a corresponding simulated microbubble image. It should be noted that in some embodiments of this specification, only a single microbubble is placed in a single simulated microbubble image.

[0143] Step 403, convolve the simulated microbubble image with a scatter plot to obtain a simulated microbubble image with a random distribution.

[0144] Among them, the scatter plot refers to the scatter image corresponding to the point spread function.

[0145] For example, taking Figure 5 the shown process of generating a simulated microbubble image as an example, by using multiple simulated microbubble images (Figure 5 the left figure in Figure 5 is convolved with the scatter plot ( Figure 5 the middle figure in

[0146] to obtain a simulated microbubble image with a random distribution (

[0147] the right figure in

[0148] Step 404: Assign microbubble velocities to the simulated microbubble image with a random distribution to obtain a simulated microbubble image in motion under the imaging parameter conditions.

[0149] In some embodiments of the present specification, the microbubble velocity assignment includes both a velocity value and a direction, both of which can be randomly selected; by randomly changing the velocity and direction of microbubble movement, the data becomes more complex and closer to real data.

[0150] Embodiments of the present specification also provide a method for training a super-resolution ultrasound imaging model. Referring to Figure 6 as shown, in some embodiments of the present specification, the method for training a super-resolution ultrasound imaging model may include the following steps:

[0151] Step 601: Obtain a microbubble image dataset; the microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator.

[0152] Among them, the microbubble image generator is obtained by training a generative network based on microbubble images. The microbubble images are converted from corresponding second ultrasound images, and the second ultrasound images are obtained by filtering background signals from corresponding third ultrasound images. The third ultrasound images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions. In other words, at least some or all of the microbubble images in the microbubble image dataset are generated based on the above microbubble image generation method.

[0153] Step 602: Use the microbubble image dataset to train a deep neural network to obtain a super-resolution ultrasound imaging model.

[0154] As Figure 7As shown, the microbubble images can be input into the deep neural network in batches to predict and output a super-resolution ultrasound image containing the position and velocity information of the microbubbles until the output of the deep neural network meets the preset conditions, and the deep neural network at this time is used as the super-resolution ultrasound imaging model.

[0155] In the embodiments of this specification, since the microbubble image generator can generate a large number of high-quality data samples, when training the deep neural network using the microbubble image dataset, a sufficient number of training data samples can be obtained; furthermore, the trained super-resolution ultrasound imaging model has better prediction accuracy and generalization ability.

[0156] The embodiments of this specification also provide a super-resolution ultrasound imaging method. Refer to Figure 8 As shown, in some embodiments of this specification, the super-resolution ultrasound imaging method may include the following steps:

[0157] Step 801, obtain multiple frames of first ultrasound images of the target object.

[0158] Among them, the multiple frames of first ultrasound images of the target object may refer to B-mode ultrasound images of human body parts.

[0159] Step 802, input the multiple frames of first ultrasound images into the super-resolution ultrasound imaging model to generate a super-resolution ultrasound image of the target object.

[0160] Among them, the super-resolution ultrasound imaging model is obtained by training a deep neural network using a microbubble image dataset. The microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator. The microbubble image generator is obtained by training a generative network based on microbubble images. The microbubble images are obtained by converting the corresponding second ultrasound images, and the second ultrasound images are obtained by filtering the background signals in the corresponding third ultrasound images. The third ultrasound images are microbubble images of biological tissues injected with microbubble contrast agents under different imaging parameter conditions. In other words, the super-resolution ultrasound imaging model is trained using the Figure 6 model training method shown.

[0161] For example, in the exemplary embodiment as shown in Figure 9 by inputting the in-vivo B-mode (i.e., the ultrasound image of the in-vivo or living body) on the left side as shown in Figure 9 into the super-resolution ultrasound imaging model for processing, the super-resolution ultrasound image as shown on the right side in Figure 9 can be correspondingly obtained.

[0162] In the embodiments of this specification, since the super-resolution ultrasound imaging model has better prediction accuracy and generalization ability, when performing super-resolution ultrasound imaging based on the super-resolution ultrasound imaging model, a more accurate super-resolution ultrasound image of the target object can be obtained.

[0163] Although the process flows described above include multiple operations that occur in a specific order, it should be clearly understood that these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel (e.g., using a parallel processor or a multi-threaded environment).

[0164] Corresponding to the above super-resolution ultrasound imaging method, an embodiment of this specification also provides a super-resolution ultrasound imaging device. Refer to Figure 10 As shown, in some embodiments of this specification, the super-resolution ultrasound imaging device may include:

[0165] A first acquisition module 11, configured to acquire multiple first ultrasound images of a target object;

[0166] A first generation module 12, configured to input the multiple first ultrasound images into a super-resolution ultrasound imaging model to generate a super-resolution ultrasound image of the target object; the super-resolution ultrasound imaging model is obtained by training a deep neural network using a microbubble image dataset, the microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on microbubble images, the microbubble images are obtained by converting corresponding second ultrasound images, and the second ultrasound images are obtained by filtering background signals from corresponding third ultrasound images, and the third ultrasound images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions.

[0167] Corresponding to the above microbubble image generation method, an embodiment of this specification also provides a microbubble image generation device. Refer to Figure 11 As shown, in some embodiments of this specification, the microbubble image generation device may include:

[0168] An acquisition module 21, configured to acquire third ultrasound images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0169] A filtering module 22, configured to filter background signals from the third ultrasound images to obtain second ultrasound images;

[0170] A conversion module 23, configured to convert the second ultrasound images into microbubble images;

[0171] A first training module 24, configured to train a generative network based on the microbubble images to obtain a microbubble image generator;

[0172] A second generation module 25, configured to generate simulated microbubble images according to the microbubble image generator.

[0173] Corresponding to the model training method of the above super-resolution ultrasound imaging model, an embodiment of this specification further provides a model training device for a super-resolution ultrasound imaging model. Refer to Figure 12 As shown in

[0174] A second acquisition module 31, configured to acquire a microbubble image dataset; the microbubble image dataset includes simulated microbubble images generated based on a microbubble image generator, the microbubble image generator is obtained by training a generative network based on microbubble images, the microbubble images are obtained by converting corresponding second ultrasound images, the second ultrasound images are obtained by filtering background signals in corresponding third ultrasound images, and the third ultrasound images are microbubble images of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions;

[0175] A second training module 32, configured to train a deep neural network by using the microbubble image dataset to obtain a super-resolution ultrasound imaging model.

[0176] For the convenience of description, when describing the above device, various units are described separately according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0177] An embodiment of this specification further provides a computer device. As Figure 13As shown, in some embodiments of this specification, the computer device 1302 may include one or more processors 1304, such as one or more central processing units (CPUs) or graphics processing units (GPUs), and each processing unit may implement one or more hardware threads. The computer device 1302 may also include any memory 1306 for storing any kind of information such as code, settings, data, etc. In a specific embodiment, a computer program stored on the memory 1306 and executable on the processor 1304, when run by the processor 1304, may execute the instructions of the method described in any of the above embodiments. Non-limitingly, for example, the memory 1306 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory devices, hard disks, optical discs, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent a fixed or removable component of the computer device 1302. In one case, when the processor 1304 executes the associated instructions stored in any memory or combination of memories, the computer device 1302 may perform any operation of the associated instructions. The computer device 1302 also includes one or more drive mechanisms 1308 for interacting with any memory, such as a hard disk drive mechanism, an optical disc drive mechanism, etc.

[0178] The computer device 1302 may also include an input / output interface 1310 (I / O) for receiving various inputs (via the input device 1312) and for providing various outputs (via the output device 1314). A specific output mechanism may include a presentation device 1316 and an associated graphical user interface 1318 (GUI). In other embodiments, the input / output interface 1310 (I / O), the input device 1312, and the output device 1314 may not be included, and it may only be a computer device in a network. The computer device 1302 may also include one or more network interfaces 1320 for exchanging data with other devices via one or more communication links 1322. One or more communication buses 1324 couple the components described above together.

[0179] The communication link 1322 may be implemented in any way, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 1322 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0180] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), computer-readable storage media, and computer program products according to some embodiments of the present specification. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processors to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processors produce a device for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0181] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processors to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device that implements the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0182] These computer program instructions can also be loaded onto a computer or other programmable data processors, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0183] In a typical configuration, a computer device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0184] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0185] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computer device. As defined in this specification, a computer-readable medium does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0186] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0187] The embodiments of this specification can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The embodiments of this specification can also be practiced in a distributed computing environment where tasks are performed by remote processors connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0188] It should also be understood that in the embodiments of this specification, the term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Additionally, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0189] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the differences between each embodiment and other embodiments are emphasized. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and reference can be made to the corresponding parts of the method embodiments for relevant details.

[0190] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0191] The above description is only for the embodiments of this application and does not limit this application. For those skilled in the art, various modifications and changes can be made to this application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included within the scope of the claims of this application.

Claims

1. A super-resolution ultrasonic imaging method, characterized in that: include: Acquire a plurality of frames of first ultrasound images of the target object; The multiple frames of first ultrasound images are input into a super-resolution ultrasound imaging model to generate a super-resolution ultrasound image of the target object; the super-resolution ultrasound imaging model is obtained by training a deep neural network using a microbubble image data set, the microbubble image data set includes a simulated microbubble image generated by a microbubble image generator, the microbubble image generator is obtained by training a generative network based on the microbubble image, the microbubble image is converted from a corresponding second ultrasound image, the second ultrasound image is obtained by filtering out a background signal in a corresponding third ultrasound image, and the third ultrasound image is a microbubble image of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions.

2. The super-resolution ultrasonic imaging method according to claim 1, characterized in that: The imaging parameter conditions include imaging frequency, the number of ultrasonic transducer array elements, the ultrasonic transducer array element width and / or imaging pulse length.

3. The super-resolution ultrasonic imaging method according to claim 1, characterized in that: The filtering out the background signal in the corresponding third ultrasonic image includes: The third ultrasonic image is filtered to filter out biological tissue signals and noise signals in the third ultrasonic image.

4. The super-resolution ultrasonic imaging method according to claim 1, characterized in that: The microbubble image is obtained by the following method: Identify a local maximum pixel point or a centroid in the third ultrasonic image; the grayscale value of the local maximum pixel point or the centroid is not less than the grayscale values ​​of the surrounding pixels; Taking out each local maximum pixel point or centroid and surrounding pixels in the third ultrasonic image to form a second ultrasonic image; Performing a correlation operation on the second ultrasonic image and the point spread function to obtain a correlation factor between the second ultrasonic image and the point spread function; Determining whether the correlation factor is greater than a correlation factor threshold; If the correlation factor is not greater than the correlation factor threshold, filtering out the second ultrasound image; If the correlation factor is greater than the correlation factor threshold, calculating the center position of each local maximum pixel point or centroid in the second ultrasound image by a microbubble positioning method; Each local maximum pixel point or centroid in the second ultrasonic image is translated to a corresponding central position to obtain a microbubble image.

5. The super-resolution ultrasonic imaging method according to claim 4, characterized in that: The method further comprises: performing a correlation operation on the second ultrasound image and a point spread function, comprising: According to the formula performing a correlation operation on the second ultrasound image and a point spread function; Wherein, c is the correlation factor, N is the number of local maximum pixel points or centroids in the second ultrasound image, S(i) is the gray value of the i-th local maximum pixel point or centroid in the second ultrasound image, G(i) is the gray value of the i-th pixel point in the point spread function, μ S is the mean gray value of the local maximum pixel point or centroid in the second ultrasound image, σ S is the standard deviation of the gray value of the local maximum pixel or centroid in the second ultrasound image, μ G is the mean gray value of the pixel in the point spread function, σ G is the standard deviation of the grayscale value of the pixel in the point spread function, and * represents convolution.

6. The super-resolution ultrasonic imaging method according to claim 4, characterized in that: The point spread function comprises: Among them, G is the point spread function, e is a natural constant, x and y are the horizontal and vertical coordinates of the pixel point in the point spread function, σ x is the lateral width of G, and d is the imaging depth, s is the number of ultrasonic transducer array elements, p is the width of the ultrasonic transducer array element, σ y is the longitudinal width of G, and σ y =l, l is the length of the imaging pulse.

7. The super-resolution ultrasonic imaging method according to claim 4, characterized in that: Extracting each local maximum pixel point or centroid and surrounding pixels in the third ultrasonic image, including: Identify whether the surrounding pixels of each local maximum pixel point or centroid in the third ultrasonic image include the local maximum pixel point or centroid; If a local maximum pixel point or a centroid is included, when each local maximum pixel point or a centroid and its surrounding pixels in the third ultrasonic image are taken out, the local maximum pixel point or a centroid among the surrounding pixels is retained.

8. The super-resolution ultrasonic imaging method according to claim 1, characterized in that: The microbubble image generator generates a simulated microbubble image in the following manner: Obtain imaging parameter conditions; Calling the microbubble image generator according to the imaging parameter conditions to generate a corresponding simulated microbubble image; Convolving the simulated microbubble image with the scatter plot to obtain a randomly distributed simulated microbubble image; Assigning microbubble velocity to the randomly distributed simulated microbubble image to obtain a simulated microbubble image in motion under the imaging parameter conditions; The above steps of generating simulated microbubble images are repeated to obtain simulated microbubble images in motion under different imaging parameter conditions.

9. A method for generating a microbubble image, characterized in that: include: Acquiring a third ultrasound image of the biological tissue injected with the microbubble contrast agent under different imaging parameter conditions; filtering out background signals in the third ultrasonic image to obtain a second ultrasonic image; converting the second ultrasound image into a microbubble image; Training a generative network based on the microbubble image to obtain a microbubble image generator; A simulated microbubble image is generated according to the microbubble image generator.

10. A model training method for a super-resolution ultrasound imaging model, characterized in that: include: Acquire a microbubble image data set; the microbubble image data set includes a simulated microbubble image generated by a microbubble image generator, the microbubble image generator is obtained by training a generative network based on the microbubble image, the microbubble image is converted from a corresponding second ultrasound image, the second ultrasound image is obtained by filtering out a background signal in a corresponding third ultrasound image, and the third ultrasound image is a microbubble image of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions; The microbubble image dataset is used to train a deep neural network to obtain a super-resolution ultrasound imaging model.

11. A super-resolution ultrasonic imaging device, characterized in that: include: A first acquisition module, used to acquire multiple frames of first ultrasound images of the target object; The first generation module is used to input the multiple frames of first ultrasound images into a super-resolution ultrasound imaging model to generate a super-resolution ultrasound image of the target object; the super-resolution ultrasound imaging model is obtained by training a deep neural network using a microbubble image data set, the microbubble image data set includes a simulated microbubble image generated by a microbubble image generator, the microbubble image generator is obtained by training a generative network based on the microbubble image, the microbubble image is converted from the corresponding second ultrasound image, the second ultrasound image is obtained by filtering out the background signal in the corresponding third ultrasound image, and the third ultrasound image is a microbubble image of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions.

12. A microbubble image generating device, characterized in that: include: An acquisition module, used for acquiring a third ultrasonic image of the biological tissue injected with the microbubble contrast agent under different imaging parameter conditions; A filtering module, used for filtering out background signals in the third ultrasonic image to obtain a second ultrasonic image; a conversion module, configured to convert the second ultrasound image into a microbubble image; A first training module, used for training a generative network based on the microbubble image to obtain a microbubble image generator; The second generating module is used to generate a simulated microbubble image according to the microbubble image generator.

13. A model training device for a super-resolution ultrasound imaging model, characterized in that: include: A second acquisition module is used to acquire a microbubble image data set; the microbubble image data set includes a simulated microbubble image generated by a microbubble image generator, the microbubble image generator is obtained by training a generative network based on the microbubble image, the microbubble image is converted from a corresponding second ultrasound image, the second ultrasound image is obtained by filtering out a background signal in a corresponding third ultrasound image, and the third ultrasound image is a microbubble image of a biological tissue injected with a microbubble contrast agent under different imaging parameter conditions; The second training module is used to train a deep neural network using the microbubble image data set to obtain a super-resolution ultrasound imaging model.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the computer program is executed by the processor, the computer program executes the instructions of the method according to any one of claims 1 to 10.

15. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor of a computer device, the computer program executes the instructions of the method according to any one of claims 1 to 10.

16. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor of a computer device, the computer program executes instructions of the method according to any one of claims 1 to 10.

Images