High-resolution real-time three-dimensional ultrasonic imaging system and method

Abstract

The invention discloses a high-resolution real-time three-dimensional ultrasonic imaging system and method, and belongs to the technical field of medical imaging. Comprising an acquisition processing module, a data registration module, a three-dimensional reconstruction module, a learning optimization module, a modeling interaction module, an intelligent extraction module, an image fusion module, a real-time processing module, a correction feedback module, a multi-dimensional analysis module and a storage sharing module. The data compression efficiency is improved, the data volume is reduced, the storage and transmission pressure is reduced, the complexity of data processing is reduced, the flexibility in different scenes is enhanced, the system is helped to adapt to complex medical environments and various clinical application requirements, efficient operation in complex and dynamic scenes can be achieved, and the system is suitable for popularization and application. According to the method, the distortion problem caused by local error accumulation is reduced, the imaging resolution is greatly improved, the image quality is improved, and the diseased region or abnormal tissue can be identified more accurately.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging technology, and in particular to a high-resolution real-time three-dimensional ultrasonic imaging system and method. Background Art

[0002] With the advancement of medical technology and the increasing demand for high-precision medical imaging, three-dimensional ultrasound imaging technology has gradually become one of the hot spots in the fields of diagnosis, treatment and research. Compared with traditional two-dimensional ultrasound, three-dimensional ultrasound can provide more comprehensive tissue structure information, especially in the imaging of complex anatomical structures. However, although three-dimensional ultrasound imaging technology has shown great potential in applications, existing technologies still face many challenges in terms of resolution, real-time performance, noise suppression, etc., especially when dealing with fast dynamic imaging. How to perform efficient processing without sacrificing image quality has become the focus of researchers.

[0003] After searching, Chinese patent number CN201910054043.0 discloses a system and method for three-dimensional ultrasound imaging. Although this invention reduces the error rate of doctors' surgical operations, improves the success rate of initial operations, and reduces doctors' surgical fatigue, it has low data compression efficiency, large data volume, increased storage and transmission pressure, and increased data processing complexity, which is not conducive to the system's adaptation to complex medical environments and various clinical application needs; in addition, the existing three-dimensional ultrasound imaging system and method cannot operate efficiently in complex and dynamic scenes, and there are many distortion problems caused by local error accumulation, which reduces the imaging resolution. For this reason, we propose a high-resolution real-time three-dimensional ultrasound imaging system and method. Summary of the invention

[0004] The purpose of the present invention is to solve the defects in the prior art and to propose a high-resolution real-time three-dimensional ultrasonic imaging system and method.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A high-resolution real-time three-dimensional ultrasonic imaging system, comprising an acquisition processing module, a data registration module, a three-dimensional reconstruction module, a learning optimization module, a modeling interaction module, an intelligent extraction module, an image fusion module, a real-time processing module, a correction feedback module, a multi-dimensional analysis module and a storage sharing module;

[0007] The acquisition and processing module is used to acquire real-time two-dimensional ultrasonic data of the detected area through a high-frequency ultrasonic probe, and perform adaptive compression and preprocessing;

[0008] The data registration module is used to register and calibrate multiple sets of two-dimensional ultrasound data;

[0009] The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on the two-dimensional ultrasound data to generate a preliminary three-dimensional ultrasound image;

[0010] The learning optimization module is used to perform denoising and enhancement on the initially reconstructed three-dimensional ultrasound image;

[0011] The modeling interaction module is used to establish an interactive virtual model for users to dynamically operate and observe the target area;

[0012] The intelligent extraction module is used to extract feature points or regions of interest from the three-dimensional ultrasound image;

[0013] The image fusion module is used to fuse different imaging data with three-dimensional ultrasound data to form a multimodal image;

[0014] The correction feedback module is used to monitor the errors in the imaging process in real time and automatically adjust the acquisition parameters;

[0015] The multidimensional analysis module is used to analyze the generated three-dimensional images and generate detailed statistical reports and diagnostic results;

[0016] The storage sharing module is used to store the three-dimensional ultrasound data and the virtual model in the cloud.

[0017] As a further solution of the present invention, the specific steps of adaptive compression of the acquisition processing module are as follows:

[0018] S1.1: Decompose the collected two-dimensional ultrasonic signal according to different frequencies or scales to form multiple scale levels, and then calculate the number of sampling points at different scale levels through an adaptive sampling density function, and take the lowest frequency level signal after scale decomposition as the root node for sampling;

[0019] S1.2: For each group of root nodes selected, obtain the child nodes containing high-frequency components corresponding to each root node according to the scale decomposition result, and then for the parent node of each layer, expand the corresponding number of child nodes according to the adaptive sampling density of the current layer;

[0020] S1.3: After each round of sub-node generation is completed, if the sub-node is generated in the high-frequency area, the positions of each sub-node are densely distributed according to the local characteristics of the signal. Otherwise, the sub-nodes are evenly distributed in various parts of the signal;

[0021] S1.4: Repeat the expansion and allocation of child nodes until the predetermined finest scale is reached and then stop expanding the nodes. At each newly generated node, the importance of each sampling point to the signal reconstruction is evaluated, and nodes with importance lower than a preset threshold are marked as redundant nodes and removed for adaptive compression. After the compression is completed, the signal is restored by multi-scale interpolation.

[0022] As a further solution of the present invention, the data registration module performs the following specific steps to register and calibrate multiple groups of two-dimensional ultrasound slices:

[0023] S2.1: collecting the registration transformation functions between the existing groups of different two-dimensional ultrasound data, collecting the transformation parameters required by each registration transformation function, and initializing the pheromone amount of each transformation parameter, and then evaluating the registration solution quality of each transformation parameter, and generating multiple groups of pathfinders;

[0024] S2.2: Each group of pathfinders randomly takes a group of preprocessed two-dimensional ultrasound data as the starting point, and calculates the selection probability of the next data and its transformation parameters according to the pheromone quantity and the quality of the registration solution. Then the pathfinder moves to the position of the two-dimensional ultrasound data and its transformation parameters with the highest selection probability;

[0025] S2.3: After a round of path selection, each pathfinder generates a complete registration scheme and calculates the total registration error of the path. According to the registration error, the corresponding path is evaluated. The smaller the path error, the better the quality of the path. At the same time, after each round of iteration, the pheromone is updated according to the path quality.

[0026] S2.4: Repeat path selection, evaluation and pheromone update until the pheromone change value reaches the preset change threshold. When the iteration stops, select the path with the smallest path error, and align each group of two-dimensional ultrasound data based on the change parameters of the path, and correct the data errors of each group after alignment.

[0027] As a further solution of the present invention, the specific steps of denoising and enhancing the learning optimization module are as follows:

[0028] S3.1: Collecting each set of historical three-dimensional ultrasound images, and inputting each set of generated historical three-dimensional ultrasound images into a generator, the generator processes each set of input three-dimensional ultrasound images through its neural network structure, and generates an enhanced image;

[0029] S3.2: The enhanced image generated by the generator and the real high-quality 3D ultrasound image are input into the discriminator. The discriminator judges each input image and outputs a score, where a score of 1 represents a real image and a score of 0 represents a generated image.

[0030] S3.3: The discriminator calculates the loss value of the discriminator based on the judgment results of the real image and the generated image, and optimizes the parameters of the discriminator through back propagation. At the same time, the generator calculates its own loss value based on the detection of the generated image of the discriminator, and optimizes the parameters of the generator through back propagation. The alternating adversarial training between the generator and the discriminator is repeated until the preset training time is reached;

[0031] S3.4: After the training is completed, the initially generated three-dimensional ultrasound image is input into the generator, and then the generator removes the noise in the initially reconstructed three-dimensional image through its neural network structure, enhances the image details, and outputs the final three-dimensional ultrasound image.

[0032] A high-resolution real-time three-dimensional ultrasonic imaging method, the specific steps of the imaging method are as follows:

[0033] Q1. Preprocess the acquired two-dimensional ultrasound data and create a preliminary three-dimensional ultrasound image;

[0034] Q2. Denoising and enhancing three-dimensional ultrasound images through GAN network;

[0035] Q3. Render and visualize 3D ultrasound images and build interactive 3D models;

[0036] Q4. Generate detailed statistical reports and diagnostic results based on 3D ultrasound images;

[0037] Q5. Store 3D ultrasound data and virtual models in the cloud.

[0038] As a further solution of the present invention, the specific steps of establishing the preliminary three-dimensional ultrasound image in Q1 are as follows:

[0039] S4.1: Collect two-dimensional ultrasound data of the same target area at different angles, initialize each set of two-dimensional ultrasound data, map the state space of the three-dimensional image into the quantum system, and represent it with the corresponding quantum state, and map each set of voxels in the state space into the corresponding superposition state of quantum bits;

[0040] S4.2: Calculate the similarity between the distribution of each voxel and the two-dimensional ultrasound data, and mark the quantum bits corresponding to the voxels whose similarity exceeds the preset threshold as the optimal solution, and then perform phase inversion processing on the marked optimal solution. After the phase inversion is completed, perform symmetric reflection on the current quantum state through quantum gate operation, and amplify the probability amplitude of the corresponding quantum state to perform diffusion processing on the target state;

[0041] S4.3: Repeat the selection, phase reversal and diffusion operations until the maximum number of iterations T is reached, where Where N represents the total number of voxel distributions. After the iteration is completed, the quantum state information of the current quantum system is measured to obtain the optimal voxel distribution. According to the optimal voxel distribution result, the three-dimensional voxel distribution that best matches the two-dimensional ultrasound data projection is found to preliminarily reconstruct the three-dimensional ultrasound image.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention decomposes the collected two-dimensional ultrasonic signal according to different frequencies or scales to form multiple scale levels, and then calculates the number of sampling points at different scale levels through an adaptive sampling density function, and samples the lowest frequency level signal after scale decomposition as the root node. For each group of root nodes screened out, the child nodes containing high-frequency components corresponding to each root node are obtained according to the scale decomposition result. Then, for the parent node of each layer, the corresponding number of child nodes is expanded according to the adaptive sampling density of the current layer. After each round of child node generation is completed, if the child node is generated in the high-frequency area, the position of each child node is densely sampled according to the local characteristics of the signal. Set distribution, otherwise, the child nodes are evenly distributed in various parts of the signal, and the expansion and allocation of child nodes are repeated until the predetermined finest scale is reached and the node expansion is stopped. At each newly generated node, the importance of each sampling point to the signal reconstruction is evaluated, and the nodes with importance lower than the preset threshold are marked as redundant nodes for removal for adaptive compression. After the compression is completed, the signal is restored through multi-scale interpolation, which improves data compression efficiency, reduces data volume, reduces storage and transmission pressure, reduces the complexity of data processing, and enhances flexibility in different scenarios, which helps the system adapt to complex medical environments and various clinical application needs.

[0044] 2. The present invention collects two-dimensional ultrasound data of the same target area at different angles, and initializes each group of two-dimensional ultrasound data, maps the state space of the three-dimensional image into the quantum system, and represents it with the corresponding quantum state, and maps each group of voxels in the state space to the quantum bit of the corresponding superposition state, calculates the similarity between each voxel distribution and the two-dimensional ultrasound data, and takes the quantum bit corresponding to the voxel whose similarity exceeds the preset threshold as the optimal solution and marks it, and then performs phase inversion processing on the marked optimal solution. After the phase inversion is completed, the current quantum state is symmetrically reflected through the quantum gate operation, and the probability amplitude of the corresponding quantum state is amplified to perform diffusion processing on the target state, and the selection, phase inversion and diffusion operations are repeated until the maximum number of iterations is reached. After the iteration is completed, the information of each quantum state in the current quantum system is measured to obtain the optimal voxel distribution. According to the optimal voxel distribution result, the three-dimensional voxel distribution that best matches the projection of the two-dimensional ultrasound data is found to preliminarily reconstruct the three-dimensional ultrasound image, which can operate efficiently in complex and dynamic scenes, reduce the distortion problem caused by local error accumulation, greatly improve the imaging resolution, improve image quality, and help to more accurately identify lesions or abnormal tissues. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0046] Figure 1A system block diagram of a high-resolution real-time three-dimensional ultrasonic imaging system proposed by the present invention;

[0047] Figure 2 This is a flowchart of a high-resolution real-time three-dimensional ultrasonic imaging method proposed by the present invention. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0049] Example 1

[0050] Reference Figure 1 A high-resolution real-time three-dimensional ultrasonic imaging system includes an acquisition and processing module, a data registration module, a three-dimensional reconstruction module, a learning optimization module, a modeling interaction module, an intelligent extraction module, an image fusion module, a real-time processing module, a correction feedback module, a multi-dimensional analysis module and a storage sharing module.

[0051] The acquisition and processing module is used to obtain real-time two-dimensional ultrasonic data of the detected area through a high-frequency ultrasonic probe, and perform adaptive compression and preprocessing.

[0052] Specifically, the collected two-dimensional ultrasonic signal is decomposed according to different frequencies or scales to form multiple scale levels, and then the number of sampling points at different scale levels is calculated through an adaptive sampling density function, and the lowest frequency level signal after scale decomposition is used as the root node for sampling. For each group of root nodes screened out, the child nodes containing high-frequency components corresponding to each root node are obtained according to the scale decomposition result. Then, for the parent node of each layer, the corresponding number of child nodes are expanded according to the adaptive sampling density of the current layer. After each round of child node generation is completed, if the child node is generated in the high-frequency area, the position of each child node is densely distributed according to the local characteristics of the signal. Otherwise, the child nodes are evenly distributed in various parts of the signal. The child nodes are repeatedly expanded and allocated until the node expansion is stopped after reaching the predetermined finest scale. At each newly generated node, the importance of each sampling point to the signal reconstruction is evaluated, and the nodes with importance lower than the preset threshold are marked as redundant nodes for removal for adaptive compression. After the compression is completed, the signal is restored by multi-scale interpolation.

[0053] The data registration module is used to register and calibrate multiple sets of two-dimensional ultrasound data.

[0054] Specifically, the registration transformation functions between the existing groups of different two-dimensional ultrasonic data are collected, the transformation parameters required for each registration transformation function are collected, and the pheromone amount of each transformation parameter is initialized at the same time. Then, the registration solution quality of each transformation parameter is evaluated, and multiple groups of pathfinders are generated. Each group of pathfinders randomly takes a group of preprocessed two-dimensional ultrasonic data as the starting point, and calculates the selection probability of the next data and its transformation parameters according to the pheromone amount and the registration solution quality. Then, the pathfinder moves to the location of the two-dimensional ultrasonic data and its transformation parameters with the highest selection probability. After a round of path selection, each pathfinder generates a complete registration scheme, and calculates the total registration error of the path. According to the registration error, the corresponding path is evaluated. The smaller the path error, the better the quality of the path. At the same time, after each round of iteration, the pheromone is updated according to the path quality, and the path selection, evaluation and pheromone update are repeated until the pheromone change value reaches the preset change threshold range. When the iteration stops, the path with the smallest path error is selected, and according to the change parameters of the path, each group of two-dimensional ultrasonic data is registered, and the data error of each group after registration is corrected.

[0055] The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on two-dimensional ultrasound data to generate a preliminary three-dimensional ultrasound image; the learning optimization module is used to denoise and enhance the preliminary reconstructed three-dimensional ultrasound image.

[0056] Specifically, each group of historical three-dimensional ultrasound images is collected, and each group of generated historical three-dimensional ultrasound images is input into the generator. The generator processes each group of input three-dimensional ultrasound images through its neural network structure and generates an enhanced image. The enhanced image generated by the generator and the real high-quality three-dimensional ultrasound image are input into the discriminator. The discriminator judges each input image and outputs a score, where a score of 1 represents a real image and a score of 0 represents a generated image. The discriminator calculates the loss value of the discriminator based on the judgment results of the real image and the generated image, and optimizes the parameters of the discriminator through back propagation. At the same time, the generator calculates its own loss value based on the detection of the generated image of the discriminator, and optimizes the parameters of the generator through back propagation. The alternating adversarial training between the generator and the discriminator is repeated until the preset training time is reached and stopped. After the training is completed, the initially generated three-dimensional ultrasound image is input into the generator. Then, the generator removes the noise in the initially reconstructed three-dimensional image through its neural network structure, enhances the image details, and outputs the final three-dimensional ultrasound image.

[0057] The modeling interaction module is used to establish an interactive virtual model for users to dynamically operate and observe the target area; the intelligent extraction module is used to extract feature points or areas of interest from three-dimensional ultrasound images; the image fusion module is used to fuse different imaging data with three-dimensional ultrasound data to form multimodal images; the correction feedback module is used to monitor errors in the imaging process in real time and automatically adjust the acquisition parameters; the multidimensional analysis module is used to analyze the generated three-dimensional images and generate detailed statistical reports and diagnostic results; the storage sharing module is used to store three-dimensional ultrasound data and virtual models in the cloud.

[0058] Example 2

[0059] Reference Figure 2 , a high-resolution real-time three-dimensional ultrasound imaging method, the imaging method has the following specific steps:

[0060] Preprocess the acquired two-dimensional ultrasound data and create a preliminary three-dimensional ultrasound image.

[0061] Specifically, two-dimensional ultrasound data of the same target area at different angles are collected, and each group of two-dimensional ultrasound data is initialized, the state space of the three-dimensional image is mapped to the quantum system and represented by the corresponding quantum state, and each group of voxels in the state space is mapped to the quantum bit of the corresponding superposition state, and the similarity between each voxel distribution and the two-dimensional ultrasound data is calculated, and the quantum bit corresponding to the voxel whose similarity exceeds the preset threshold is taken as the optimal solution and marked, and then the marked optimal solution is phase-reversed. After the phase reversal is completed, the current quantum state is symmetrically reflected through the quantum gate operation, and the probability amplitude of the corresponding quantum state is amplified to diffuse the target state, and the selection, phase reversal and diffusion operations are repeated until the maximum number of iterations T is reached, where Where N represents the total number of voxel distributions. After the iteration is completed, the quantum state information of the current quantum system is measured to obtain the optimal voxel distribution. According to the optimal voxel distribution result, the three-dimensional voxel distribution that best matches the two-dimensional ultrasound data projection is found to preliminarily reconstruct the three-dimensional ultrasound image.

[0062] Denoising and enhancement of three-dimensional ultrasound images are performed using the GAN network.

[0063] Render and visualize 3D ultrasound images and create interactive 3D models.

[0064] Generate detailed statistical reports and diagnostic results based on 3D ultrasound images.

[0065] Store 3D ultrasound data and virtual models in the cloud.

Claims

1. A high-resolution real-time three-dimensional ultrasonic imaging system, characterized in that: It includes acquisition and processing module, data registration module, 3D reconstruction module, learning optimization module, modeling interaction module, intelligent extraction module, image fusion module, real-time processing module, correction feedback module, multi-dimensional analysis module and storage sharing module; The acquisition and processing module is used to acquire real-time two-dimensional ultrasonic data of the detected area through a high-frequency ultrasonic probe, and perform adaptive compression and preprocessing; The data registration module is used to register and calibrate multiple sets of two-dimensional ultrasound data; The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on the two-dimensional ultrasound data to generate a preliminary three-dimensional ultrasound image; The learning optimization module is used to perform denoising and enhancement on the initially reconstructed three-dimensional ultrasound image; The modeling interaction module is used to establish an interactive virtual model for users to dynamically operate and observe the target area; The intelligent extraction module is used to extract feature points or regions of interest from the three-dimensional ultrasound image; The image fusion module is used to fuse different imaging data with three-dimensional ultrasound data to form a multimodal image; The correction feedback module is used to monitor the errors in the imaging process in real time and automatically adjust the acquisition parameters; The multidimensional analysis module is used to analyze the generated three-dimensional images and generate detailed statistical reports and diagnostic results; The storage sharing module is used to store the three-dimensional ultrasound data and the virtual model in the cloud.

2. A high-resolution real-time three-dimensional ultrasonic imaging system according to claim 1, characterized in that: The specific steps of adaptive compression of the acquisition processing module are as follows: S1.1: Decompose the collected two-dimensional ultrasonic signal according to different frequencies or scales to form multiple scale levels, and then calculate the number of sampling points at different scale levels through an adaptive sampling density function, and take the lowest frequency level signal after scale decomposition as the root node for sampling; S1.2: For each group of root nodes selected, obtain the child nodes containing high-frequency components corresponding to each root node according to the scale decomposition result, and then for the parent node of each layer, expand the corresponding number of child nodes according to the adaptive sampling density of the current layer; S1.3: After each round of sub-node generation is completed, if the sub-node is generated in the high-frequency area, the positions of each sub-node are densely distributed according to the local characteristics of the signal. Otherwise, the sub-nodes are evenly distributed in various parts of the signal; S1.4: Repeat the expansion and allocation of child nodes until the predetermined finest scale is reached and then stop expanding the nodes. At each newly generated node, the importance of each sampling point to the signal reconstruction is evaluated, and nodes with importance lower than a preset threshold are marked as redundant nodes and removed for adaptive compression. After the compression is completed, the signal is restored by multi-scale interpolation.

3. A high-resolution real-time three-dimensional ultrasonic imaging system according to claim 2, characterized in that: The specific steps of the data registration module registering and calibrating multiple groups of two-dimensional ultrasound slices are as follows: S2.1: collecting the registration transformation functions between the existing groups of different two-dimensional ultrasound data, collecting the transformation parameters required by each registration transformation function, and initializing the pheromone amount of each transformation parameter, and then evaluating the registration solution quality of each transformation parameter, and generating multiple groups of pathfinders; S2.2: Each group of pathfinders randomly takes a group of preprocessed two-dimensional ultrasound data as the starting point, and calculates the selection probability of the next data and its transformation parameters according to the pheromone quantity and the quality of the registration solution. Then the pathfinder moves to the position of the two-dimensional ultrasound data and its transformation parameters with the highest selection probability; S2.3: After a round of path selection, each pathfinder generates a complete registration scheme and calculates the total registration error of the path. According to the registration error, the corresponding path is evaluated. The smaller the path error, the better the quality of the path. At the same time, after each round of iteration, the pheromone is updated according to the path quality. S2.4: Repeat path selection, evaluation and pheromone update until the pheromone change value reaches the preset change threshold. When the iteration stops, select the path with the smallest path error, and align each group of two-dimensional ultrasound data based on the change parameters of the path, and correct the data errors of each group after alignment.

4. A high-resolution real-time three-dimensional ultrasonic imaging system according to claim 1, characterized in that: The specific steps of denoising and enhancement of the learning optimization module are as follows: S3.1: Collecting each set of historical three-dimensional ultrasound images, and inputting each set of generated historical three-dimensional ultrasound images into a generator, the generator processes each set of input three-dimensional ultrasound images through its neural network structure, and generates an enhanced image; S3.2: The enhanced image generated by the generator and the real high-quality 3D ultrasound image are input into the discriminator. The discriminator judges each input image and outputs a score, where a score of 1 represents a real image and a score of 0 represents a generated image. S3.3: The discriminator calculates the loss value of the discriminator based on the judgment results of the real image and the generated image, and optimizes the parameters of the discriminator through back propagation. At the same time, the generator calculates its own loss value based on the detection of the generated image of the discriminator, and optimizes the parameters of the generator through back propagation. The alternating adversarial training between the generator and the discriminator is repeated until the preset training time is reached; S3.4: After the training is completed, the initially generated three-dimensional ultrasound image is input into the generator, and then the generator removes the noise in the initially reconstructed three-dimensional image through its neural network structure, enhances the image details, and outputs the final three-dimensional ultrasound image.

5. A high-resolution real-time three-dimensional ultrasonic imaging method, used to implement the functions of the high-resolution real-time three-dimensional ultrasonic imaging system according to any one of claims 1 to 4, characterized in that: The specific steps of the imaging method are as follows: Q1. Preprocess the acquired two-dimensional ultrasound data and create a preliminary three-dimensional ultrasound image; Q2. Denoising and enhancing three-dimensional ultrasound images through GAN network; Q3. Render and visualize 3D ultrasound images and build interactive 3D models; Q4. Generate detailed statistical reports and diagnostic results based on 3D ultrasound images; Q5. Store 3D ultrasound data and virtual models in the cloud.

6. A high-resolution real-time three-dimensional ultrasonic imaging method according to claim 5, characterized in that: The specific steps for establishing the preliminary three-dimensional ultrasound image described in Q1 are as follows: S4.1: Collect two-dimensional ultrasound data of the same target area at different angles, initialize each set of two-dimensional ultrasound data, map the state space of the three-dimensional image into the quantum system, and represent it with the corresponding quantum state, and map each set of voxels in the state space into the corresponding superposition state of quantum bits; S4.2: Calculate the similarity between the distribution of each voxel and the two-dimensional ultrasound data, and mark the quantum bits corresponding to the voxels whose similarity exceeds the preset threshold as the optimal solution, and then perform phase inversion processing on the marked optimal solution. After the phase inversion is completed, perform symmetric reflection on the current quantum state through quantum gate operation, and amplify the probability amplitude of the corresponding quantum state to perform diffusion processing on the target state; S4.3: Repeat the selection, phase reversal and diffusion operations until the maximum number of iterations T is reached, where Where N represents the total number of voxel distributions. After the iteration is completed, the quantum state information of the current quantum system is measured to obtain the optimal voxel distribution. According to the optimal voxel distribution result, the three-dimensional voxel distribution that best matches the two-dimensional ultrasound data projection is found to preliminarily reconstruct the three-dimensional ultrasound image.

Images