CBCT pseudo nuclear magnetic image reconstruction system and method based on artificial intelligence

Abstract

The invention discloses a CBCT pseudo nuclear magnetic image reconstruction system and method based on artificial intelligence. The system comprises a data acquisition and preprocessing module, a pseudo nuclear magnetic image generation module and a central control module. By introducing a deep learning technology, the system analyzes the preprocessed cone beam computed tomography image data to generate a pseudo nuclear magnetic resonance image with high resolution and high contrast. The generated pseudo nuclear magnetic image can clearly display the structural features of soft tissues and joints, significantly improves the precision of temporomandibular joint health assessment, and significantly optimizes the clinical diagnosis process and efficiency at the same time.

Description

Technical Field

[0001] The present invention relates to the field of oral orthodontics, and in particular to an orthodontic auxiliary treatment system and method based on artificial intelligence CBCT reconstruction of pseudo nuclear magnetic resonance. Background Art

[0002] In orthodontic treatment, accurate imaging data is essential for diagnosis and treatment planning. Traditional imaging methods such as CBCT (cone beam computed tomography) are widely used in the field of orthodontics. CBCT uses X-rays to perform three-dimensional imaging of the oral cavity and provide high-resolution information on the structure of teeth and bones. However, CBCT has obvious deficiencies in displaying oral soft tissues, and the contrast of soft tissues is low, which limits its application in some complex orthodontic treatments.

[0003] In contrast, magnetic resonance imaging (MRI) technology has significant advantages in soft tissue contrast, and can provide clearer soft tissue images, helping doctors better understand the soft tissue conditions inside the mouth. However, MRI equipment is expensive, complicated to operate, and takes a long time to examine, which limits its application in conventional orthodontic treatment. Therefore, how to combine the advantages of CBCT and MRI to provide oral images with both high resolution and high soft tissue contrast has become an important research direction in the current field of orthodontics.

[0004] However, the shortcomings of the prior art are:

[0005] 1. Disadvantages of CBCT in the field of orthodontics:

[0006] 1) Low soft tissue contrast: CBCT has poor display effect on soft tissue and cannot clearly distinguish soft tissue structure, which affects the accuracy of orthodontic diagnosis and treatment planning.

[0007] 2) Radiation dose issue: Although the radiation dose of CBCT is lower than that of traditional CT, there is still a certain radiation risk when it is used frequently, especially for young patients and when multiple examinations are required.

[0008] 2. Disadvantages of MRI in the field of orthodontics:

[0009] 1) Expensive equipment: MRI equipment is expensive and requires high maintenance, making it unsuitable for widespread promotion in routine orthodontic treatment.

[0010] 2) Complex operation: MRI examination takes a long time, requires high cooperation from patients, and also requires high technical skills from operators.

[0011] 3) Magnetic field restrictions: During MRI examination, patients cannot carry metal objects, and some patients (such as those with pacemakers) are not suitable for MRI examination.

[0012] 3. Deficiencies of existing image processing technology in the field of orthodontics:

[0013] Difficulty in technical integration: Current image processing technology cannot effectively combine the advantages of CBCT and MRI and cannot provide image data with high resolution and high soft tissue contrast at the same time, which limits the accuracy and effectiveness of orthodontic diagnosis and treatment. Summary of the invention

[0014] The present invention aims to solve the technical problem of insufficient soft tissue contrast of CBCT images in the prior art, and provide a system and method for reconstructing pseudo MRI images from CBCT based on artificial intelligence. The system uses a variety of deep learning models, including but not limited to the diffusion model MC-IDDPM, boundary condition diffusion model (BBDM) and conditional generative adversarial network (cGAN), to realize the processing of CBCT image data and the generation of pseudo images.

[0015] <Technical solution>

[0016] The present invention is achieved through the following technical solutions:

[0017] 1. The data acquisition and preprocessing module normalizes, reduces noise and converts the format of CBCT image data, laying the foundation for the processing of deep learning models;

[0018] 2. The pseudo-NMR image generation module generates pseudo-NMR images based on the preprocessed image data using a multi-model collaborative deep learning algorithm;

[0019] 3. The central control module coordinates the task execution between modules, supports dynamic model selection to adapt to different scenario requirements, and provides a user interaction interface for result display and parameter adjustment.

[0020] Through the above scheme, the present invention can generate pseudo-MRI images with high resolution and excellent contrast, significantly improving the accuracy of temporomandibular joint health assessment, while effectively improving clinical diagnosis efficiency.

[0021] Specifically, the present invention provides a system for reconstructing pseudo nuclear magnetic resonance images using CBCT based on artificial intelligence, comprising:

[0022] Data acquisition and preprocessing module, used to acquire cone beam computed tomography (CBCT) image data and normalize, reduce noise and convert the data format;

[0023] The pseudo-MRI image generation module extracts features from pre-processed CBCT image data based on a deep learning model to generate pseudo images with MRI characteristics.

[0024] Preferably, the system further comprises:

[0025] The central control module is used to coordinate the task execution and data transmission of each module, and provides a user interface for model selection, parameter setting and result display.

[0026] Preferably, the pseudo nuclear magnetic resonance image generation module generates pseudo nuclear magnetic resonance images using a deep learning model, and the model includes but is not limited to any one of the following:

[0027] Diffusion model MC-IDDPM, which generates high-resolution images through iterative denoising;

[0028] Boundary Condition Diffusion Model (BBDM), which enhances image clarity and contrast through boundary guidance;

[0029] Conditional Generative Adversarial Networks (cGANs), which generate fake images based on input images and conditions.

[0030] Preferably, the pseudo-NMR image generation module supports dynamic switching of models to adapt to the requirements of different scenarios for image generation speed, quality and resource requirements.

[0031] Preferably, the data acquisition and preprocessing module includes:

[0032] Normalization submodule, used to adjust the pixel value range of image data;

[0033] Noise reduction submodule, using adaptive filtering algorithm to reduce noise interference;

[0034] The format conversion submodule is used to convert the image data into the input format required by the deep learning network.

[0035] Preferably, the central control module includes:

[0036] ·Parameter management unit, used to set the model parameters of the pseudo-MRI image generation module;

[0037] User interaction unit, used to provide real-time preview and adjustment interface of image generation results.

[0038] Preferably, the characteristics of the pseudo-NMR image include:

[0039] High resolution, with a resolution of no less than 512×512 pixels;

[0040] High contrast, the contrast is not less than 256 gray levels;

[0041] ·Can clearly display the soft tissue and bone structure of the temporomandibular joint.

[0042] According to the present invention, a method for generating a pseudo nuclear magnetic resonance image based on artificial intelligence is also provided, which comprises the following steps:

[0043] Preprocess CBCT image data by normalization, noise reduction and format conversion;

[0044] Use deep learning models to generate pseudo images with MRI characteristics;

[0045] Combined with the characteristics of the artifacts, the generated results are evaluated and adjusted through the quality optimization algorithm.

[0046] Preferably, the process of generating a pseudo image by the deep learning model includes:

[0047] Using a diffusion model to generate images through iterative denoising;

[0048] Use conditional generative adversarial networks (cGANs) to generate fake images based on input images and conditions;

[0049] · Optimize the details and clarity of the generated image by combining boundary conditions.

[0050] Preferably, the quality optimization includes using perceptual loss (Perceptua l Loss), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR) to comprehensively evaluate and adjust the pseudo image.

[0051] Preferably, the pseudo-NMR image generation process supports the deployment of a cloud computing platform to improve processing efficiency and achieve remote optimization and updating of model parameters.

[0052] <Technical Effects>

[0053] 1. Improve image quality: Pseudo-MRI images are close to actual MRI images in soft tissue contrast and resolution, with a resolution of no less than 512×512 pixels and a contrast of no less than 256 gray levels;

[0054] 2. Enhanced flexibility: Supports dynamic switching of diffusion models MC-IDDPM, BBDM and cGAN, and can optimize image generation according to different clinical needs;

[0055] 3. Improve diagnostic efficiency: By combining cloud computing technology, the efficiency of image generation is increased by more than 50%, making it suitable for large-scale clinical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] A more complete understanding of the present invention and its attendant advantages and features will be more readily appreciated by reference to the following detailed description taken in conjunction with the accompanying drawings, in which:

[0057] Figure 1 The functional block diagram of the orthodontic auxiliary treatment system based on artificial intelligence CBCT reconstruction of pseudo nuclear magnetic resonance according to a preferred embodiment of the present invention is schematically shown.

[0058] Figure 2 A real CBCT image is shown.

[0059] Figure 3 A reference MR is shown.

[0060] Figure 4 The pseudo MR obtained by the present invention is shown.

[0061] It should be noted that the drawings are used to illustrate the present invention, rather than to limit the present invention. Note that the drawings showing the structures may not be drawn to scale. In addition, in the drawings, the same or similar elements are marked with the same or similar reference numerals. DETAILED DESCRIPTION

[0062] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0064] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0065] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0066] The present invention discloses an orthodontic auxiliary treatment system for reconstructing pseudo-nuclear magnetic resonance imaging (PNE) based on artificial intelligence (AI). The system converts CBCT image data into pseudo-nuclear magnetic resonance imaging (PNE) images with MRI characteristics through artificial intelligence technology, thereby combining the high resolution of CBCT and the superior soft tissue contrast of MRI, and uses artificial intelligence to convert CBCT data into MRI data to solve the problem of soft tissue clarity, providing more accurate image information for orthodontic diagnosis and treatment planning. The system includes the following modules: data acquisition and preprocessing module, CBCT image processing and pseudo-nuclear magnetic resonance imaging (PNE) reconstruction module, tooth segmentation and classification numbering module, data analysis module, orthodontic path planning module, orthodontic scheme determination module, simulation module, orthodontic scheme evaluation module, data storage module, and central control module. The system reconstructs CBCT data into pseudo-nuclear magnetic resonance imaging (PNE) images through AI technology, generates a high-precision three-dimensional oral model, thereby improving the accuracy and efficiency of orthodontic treatment and shortening the treatment cycle.

[0067] The innovation of the present invention lies in the innovative idea of ​​generating pseudo-MRI images based on CBCT:

[0068] - Image data reconstruction: By processing CBCT image data, pseudo images with MRI characteristics are generated, so that the images have both the high resolution of CBCT and the ability to display soft tissue details.

[0069] -Application of pseudo-MRI images: The generated pseudo-MRI images are applied to the diagnosis and treatment of orthodontics, providing clearer images of the internal structure of the oral cavity and assisting doctors in formulating more accurate treatment plans.

[0070] -Any method or idea to generate pseudo-MRI images using CBCT image data in orthodontic diagnosis and treatment.

[0071] -Specific applications of pseudo-MRI in orthodontics, including for diagnosis, treatment planning and evaluation.

[0072] Specifically, Figure 1 The functional block diagram of the orthodontic auxiliary treatment system based on artificial intelligence CBCT reconstruction of pseudo nuclear magnetic resonance according to a preferred embodiment of the present invention is schematically shown. Figure 1 As shown, the orthodontic auxiliary treatment system based on artificial intelligence CBCT reconstruction pseudo nuclear magnetic resonance according to the preferred embodiment of the present invention includes:

[0073] Data acquisition and preprocessing 10 is used to acquire the user's oral three-dimensional image data through the CBCT device and preprocess the acquired original CBCT data.

[0074] Preferably, the preprocessing includes: performing registration and cropping (registration, correction, splicing and cropping) on ​​the acquired original CBCT data with reference to the MRI data format.

[0075] Preferably, the preprocessing includes: performing preprocessing such as denoising and correction on the collected image data to ensure data quality. This module includes a variety of image processing algorithms to improve the clarity and consistency of the image.

[0076] Preferably, data collection and preprocessing also collects basic information of the user, including age, gender, oral health status, etc., for subsequent data analysis and processing.

[0077] Preferably, data acquisition and preprocessing perform image correction and standardization on the registered and cropped data to ensure the accuracy and consistency of the image and provide reliable basic data for subsequent image processing and analysis.

[0078] In a specific embodiment, for example, the data acquisition and preprocessing module 10 includes:

[0079] · A normalization submodule 11, used to adjust the pixel value range of the image data;

[0080] · The noise reduction submodule 12 uses an adaptive filtering algorithm to reduce noise interference;

[0081] · The format conversion submodule 13 is used to convert the image data into the input format required by the deep learning network.

[0082] The pseudo MRI image generation module 20 is used to convert the pre-processed CBCT image data into pseudo MRI image data using a deep learning model, that is, to generate MRI images based on the pre-processed CBCT image data.

[0083] Specifically, CBCT can provide high-resolution information on tooth and bone structure, but has poor display effect on soft tissue; the present invention uses a deep learning model to convert CBCT data into pseudo-MRI images, thereby improving the image's expressiveness for soft tissue.

[0084] For deep learning model selection, for example, the following specific methods can be used:

[0085] -Choose a deep learning model based on the neural network DDPM model (preferably, the latest Diffusion + Transfer model) because of its high accuracy and performance in image recognition and processing.

[0086] -The deep learning model inputs CBCT images and outputs pseudo images with MRI characteristics. The input and output structures of the model are designed as follows:

[0087] -Input: pre-processed CBCT image data.

[0088] -Output: Pseudo MRI image data.

[0089] Deep learning models include but are not limited to any of the following:

[0090] Diffusion model MC-IDDPM, which generates high-resolution images through iterative denoising;

[0091] Boundary Condition Diffusion Model (BBDM), which enhances image clarity and contrast through boundary guidance;

[0092] Conditional Generative Adversarial Networks (cGANs), which generate fake images based on input images and conditions.

[0093] In addition, for example, the pseudo-MRI image generation module supports dynamic switching of models to adapt to the requirements of different scenarios for image generation speed, quality and resource requirements.

[0094] For the training of deep learning models, for example, the following specific methods can be used:

[0095] - Use a large number of CBCT and real MRI images to train the model and establish a mapping relationship from CBCT to MRI. The training dataset should include multiple types of oral images to ensure the generalization ability of the model.

[0096] -Use data augmentation techniques (such as rotation, flipping, scaling, etc.) to increase the diversity of training data and improve the robustness of the model.

[0097] - Use cross-validation method to evaluate model performance, adjust model parameters, and optimize model effects.

[0098] For example, the process of generating artificial images by a deep learning model may include:

[0099] · Using a diffusion model, images are generated through iterative denoising;

[0100] · Use conditional generative adversarial networks (cGAN) to generate fake images based on input images and conditions;

[0101] · The generated image is optimized in combination with boundary conditions, such as optimizing the details and clarity of the generated image.

[0102] During the image reconstruction process, through continuous iterative training, the generated pseudo MRI images can approach the characteristics of real MRI images, with both the high resolution of CBCT and the ability to clearly display soft tissue structures.

[0103] Therefore, the characteristics of pseudo-NMR imaging that can be achieved by the present invention include:

[0104] High resolution, with a resolution of no less than 512×512 pixels;

[0105] High contrast, the contrast is not less than 256 gray levels;

[0106] ·Can clearly display the soft tissue and bone structure of the temporomandibular joint.

[0107] For example, refer to Figures 2 to 4 As shown, Figure 2 shows a real CBCT image, Figure 3 The reference MR is shown, Figure 4 The pseudo MR obtained by the present invention is shown.

[0108] Furthermore, preferably, Figure 1 As shown, the orthodontic auxiliary treatment system based on artificial intelligence CBCT reconstruction pseudo nuclear magnetic resonance according to the preferred embodiment of the present invention also includes:

[0109] The central control module 30 is used to control the operation of each module and ensure the coordination and accuracy of data processing. The central control module is responsible for managing the workflow of the system and coordinating the interaction and data transmission of each functional module.

[0110] For example, in a specific embodiment, preferably, the central control module 30 includes:

[0111] · A parameter management unit 31, used to set the model parameters of the pseudo nuclear magnetic resonance image generation module;

[0112] The user interaction unit 32 is used to provide a real-time preview and adjustment interface of the image generation result.

[0113] Accordingly, according to another preferred embodiment of the present invention, a method for generating a pseudo nuclear magnetic resonance image based on artificial intelligence is provided, which comprises the following steps:

[0114] Preprocess CBCT image data by normalization, noise reduction and format conversion;

[0115] Use deep learning models to generate artificial images with MRI characteristics;

[0116] Combined with the characteristics of the artifacts, the generated results are evaluated and adjusted through the quality optimization algorithm.

[0117] Preferably, the process of generating a pseudo image by the deep learning model includes:

[0118] Using a diffusion model to generate images through iterative denoising;

[0119] Use conditional generative adversarial networks (cGANs) to generate fake images based on input images and conditions;

[0120] Optimize the generated image by combining boundary conditions, such as optimizing the details and clarity of the generated image.

[0121] Preferably, the quality optimization includes using perceptual loss (Perceptua l Loss), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR) to comprehensively evaluate and adjust the pseudo image.

[0122] Preferably, the pseudo-NMR image generation process supports the deployment of a cloud computing platform to improve processing efficiency and achieve remote optimization and updating of model parameters.

[0123] <Optimization of pseudo-MRI images>

[0124] Fusion and enhancement can be used to optimize pseudo-MRI images, for example:

[0125] -Combining multiple image processing technologies, the contrast and clarity of pseudo-MRI images are further enhanced to ensure that key details are fully displayed.

[0126] - Through image fusion technology, multimodal imaging data is fused to generate more informative comprehensive images, thereby improving the accuracy of diagnosis and treatment.

[0127] <Verification of pseudo-MRI images>

[0128] Validation of pseudo-MRI images can be performed, for example:

[0129] -The generated pseudo-MRI images are verified and optimized multiple times, and the model parameters are adjusted by comparing them with real MRI images to improve the accuracy and reliability of the images.

[0130] - Use evaluation indicators (such as peak signal-to-noise ratio, structural similarity index, etc.) to assess the quality of pseudo-MRI images to ensure that they meet the requirements of clinical application.

[0131] <Variation plan>

[0132] The possible changes and variations of the technical solution of the present invention mainly focus on image processing technology, data fusion method, hardware optimization and improvement of processing algorithm. The following are specific design changes and variations:

[0133] 1. Alternative image processing techniques

[0134] Change direction:

[0135] -Although the present invention mainly uses deep learning technology to achieve image reconstruction, other image processing technologies can also be used to generate pseudo nuclear magnetic resonance images as long as similar effects can be achieved. For example, solutions based on traditional image processing algorithms or hybrid algorithms can also be considered.

[0136] Variations:

[0137] -Use rule-based methods for image processing and reconstruction, such as image segmentation algorithms, image enhancement algorithms, etc.

[0138] -Use other machine learning algorithms, such as support vector machine (SVM) or random forest (RF), to process CBCT image data.

[0139] 2. Multimodal Data Fusion

[0140] Change direction:

[0141] -The technical solution of the present invention can be extended to combine other types of image data, and further improve the image quality and diagnostic accuracy through multimodal data fusion. This method can provide more comprehensive diagnostic information and enhance the clinical application effect.

[0142] Variations:

[0143] - Fusion of 2D X-ray images with CBCT image data to generate comprehensive images so that doctors can obtain more comprehensive information in orthodontic diagnosis.

[0144] - Combined with ultrasound image data, using its good imaging characteristics for soft tissue, the soft tissue contrast of pseudo MRI images is enhanced. Due to the good imaging characteristics of ultrasound image data for soft tissue, when it is input into the deep learning model based on the neural network DDPM model together with the pre-processed CBCT image data, excellent pseudo MRI image data can be obtained.

[0145] 3. Hardware Acceleration

[0146] Change direction:

[0147] -Specialized hardware accelerators or computing platforms can be used to improve image processing speed and efficiency to meet the needs of clinical applications of different scales. Hardware optimization helps speed up the image processing process and meet the needs of real-time processing.

[0148] Variations:

[0149] -Develop and use specialized hardware such as Graphics Processing Units (GPUs) or Tensor Processing Units (TPUs) for image processing.

[0150] -Use field programmable gate array (FPGA) technology and customized hardware accelerator to achieve efficient image reconstruction and processing.

[0151] 4. Improved image processing algorithm

[0152] Change direction:

[0153] -Introduce new image enhancement, noise reduction and correction algorithms at various stages of image processing to further improve image quality and processing efficiency. These improvements can be achieved in image preprocessing, feature extraction and image reconstruction.

[0154] Variations:

[0155] -Use more advanced noise reduction algorithms, such as deep learning-based noise reduction networks, to improve the quality of CBCT images.

[0156] -Introducing the latest image enhancement technology to enhance image details and contrast and improve the clarity of pseudo-MRI images.

[0157] -Use more precise image correction algorithm to ensure the consistency and accuracy of image data.

[0158] 5. Flexible configuration of system modules

[0159] Change direction:

[0160] -The configuration of system modules can be adjusted according to specific application requirements to ensure the flexibility and adaptability of the system. Different clinical applications may require different module combinations and functional implementations.

[0161] Variations:

[0162] -Adjust the configuration of system modules for different clinical application scenarios, such as adding or reducing certain functional modules to meet specific needs.

[0163] -Provide a modular system architecture so that each functional module can be upgraded and replaced independently, improving the scalability and maintainability of the system.

[0164] <Other Improvements>

[0165] Preferably, the present invention can also improve the automation level of image processing and analysis by introducing more automation technologies, reduce manual intervention, and thus improve work efficiency and diagnostic accuracy. Implementation methods include, for example:

[0166] -Develop automated image analysis tools that can automatically identify and annotate key anatomical structures, such as teeth, jaws, and soft tissue lesions.

[0167] -Integrated intelligent diagnostic system combines imaging data and medical record information to automatically generate diagnostic reports and treatment recommendations to assist doctors in making decisions.

[0168] Preferably, the present invention can also provide for the formulation of personalized treatment plans. Specifically, based on pseudo-MRI data and combined with the specific conditions of the patient, a personalized treatment plan is formulated to improve the treatment effect and patient satisfaction. The implementation methods include, for example:

[0169] -Use machine learning and data analysis technology to analyze large amounts of patient data, summarize the optimal treatment path and plan, and apply it to individualized treatment.

[0170] - Develop a patient characteristics database to recommend personalized treatment plans based on the characteristics and conditions of different patients to ensure that each patient receives the most suitable treatment.

[0171] The present invention can also be used for multidisciplinary cooperation and integrated application, promote multidisciplinary cooperation, apply the technical solution of the present invention to a comprehensive diagnosis and treatment platform, and integrate the technology and knowledge of multiple disciplines such as imaging, stomatology, and computer science. For example, the implementation methods include:

[0172] -Establish a multidisciplinary research team, combining expertise from various fields to jointly develop and improve the technical solutions of the present invention.

[0173] -Integrate a variety of medical imaging and diagnosis and treatment technologies, build a comprehensive diagnosis and treatment platform, provide one-stop medical services, and improve diagnosis and treatment efficiency and effectiveness.

[0174] The present invention can also achieve data security and privacy protection, ensure the security and privacy protection of patient data during the processing and storage of image data, comply with relevant laws and regulations, and build a safe and reliable data management system. Implementation methods include, for example:

[0175] -Adopt advanced data encryption technology to ensure the security of image data during transmission and storage.

[0176] -Establish a strict data access control mechanism so that only authorized personnel can access patients’ imaging data and medical information.

[0177] -Perform data security audits and assessments regularly to promptly identify and fix potential security vulnerabilities and ensure data security and privacy protection.

[0178] <Other Applications>

[0179] The technical solution of the present invention is not limited to the field of orthodontics, but can also be extended to other medical fields that require high-quality soft tissue images, such as craniofacial surgery, otolaryngology, neurology, etc. These fields also require high-resolution and high-contrast image data to assist diagnosis and treatment.

[0180] Application examples:

[0181] -In craniofacial surgery, pseudo-MRI can be used for preoperative planning and postoperative evaluation, providing clear images of soft tissue and bone structures.

[0182] -In otolaryngology, pseudo-MRI can be used to more accurately diagnose and treat various complex soft tissue lesions.

[0183] <Technical Effects>

[0184] The present invention improves the accuracy and efficiency of orthodontic treatment through the following aspects:

[0185] 1. Improve image quality: Pseudo-MRI images combine the high resolution of CBCT and the high soft tissue contrast of MRI, making the images clearer and facilitating doctors to conduct detailed analysis of oral structures.

[0186] 2. Reduce costs and complexity: AI technology is used to generate pseudo MRI images based on CBCT images, avoiding expensive and complex MRI examinations, thereby reducing examination costs and operational complexity.

[0187] 3. Improve diagnostic accuracy: Through high-quality pseudo-MRI images, doctors can diagnose and plan orthodontic treatment more accurately, improving treatment outcomes.

[0188] 4. Shorten the treatment cycle: Automated image processing and reconstruction processes improve the efficiency of image data acquisition and processing, thereby shortening the time for diagnosis and treatment planning.

[0189] The present invention significantly improves the soft tissue representation ability of images by reconstructing CBCT data into pseudo-MRI images, thereby providing more accurate diagnosis and planning during orthodontic treatment, greatly improving treatment effectiveness and efficiency, and has important clinical application value.

[0190] In summary, the present invention provides an orthodontic auxiliary treatment system based on artificial intelligence (AI) CBCT reconstruction of pseudo-MRI, in which CBCT image data is processed by artificial intelligence technology and reconstructed into a pseudo image with MRI characteristics, and applied to the diagnosis and treatment of oral orthodontics, thereby significantly improving the image quality and diagnostic accuracy, and providing more effective assistance for orthodontic treatment.

[0191] In addition, it should be noted that, unless otherwise specified, the terms "first", "second", "third", etc. in the specification are only used to distinguish the various components, elements, steps, etc. in the specification, and are not used to indicate the logical relationship or sequential relationship between the various components, elements, steps, etc.

[0192] It is to be understood that, although the present invention has been disclosed as a preferred embodiment, the above embodiment is not intended to limit the present invention. For any person skilled in the art, without departing from the scope of the technical solution of the present invention, the technical content disclosed above can be used to make many possible changes and modifications to the technical solution of the present invention, or modified into equivalent embodiments of equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of protection of the technical solution of the present invention.

Claims

1. A system for reconstructing pseudo-nuclear magnetic resonance images using CBCT based on artificial intelligence, characterized in that: include: · Data acquisition and preprocessing module, used to acquire cone beam computed tomography (CBCT) image data and perform normalization, noise reduction and format conversion on the data; · The pseudo MRI image generation module extracts features from the preprocessed CBCT image data based on a deep learning model to generate pseudo images with MRI characteristics.

2. The system according to claim 1, characterized in that Also includes: · The central control module is used to coordinate the task execution and data transmission of each module, and provides a user interface for model selection, parameter setting and result display.

3. The system according to claim 1, characterized in that The pseudo-NMR image generation module generates pseudo-NMR images using a deep learning model, and the model includes but is not limited to any of the following: · The diffusion model MC-IDDPM generates high-resolution images through iterative denoising; · Boundary Condition Diffusion Model (BBDM), which enhances image clarity and contrast through boundary guidance; · Conditional Generative Adversarial Network (cGAN), generates fake images based on input images and conditions.

4. The system according to claim 3, characterized in that The pseudo-NMR image generation module supports dynamic switching of models to adapt to the requirements of different scenarios for image generation speed, quality and resource requirements.

5. The system according to claim 1, characterized in that The data acquisition and preprocessing module includes: · A normalization submodule is used to adjust the pixel value range of the image data; · The noise reduction submodule uses an adaptive filtering algorithm to reduce noise interference; · The format conversion submodule is used to convert image data into the input format required by the deep learning network.

6. The system according to claim 1, characterized in that The central control module comprises: · A parameter management unit, used to set the model parameters of the pseudo-MRI image generation module; · The user interaction unit is used to provide a real-time preview and adjustment interface for the image generation results.

7. The system according to claim 1, characterized in that The characteristics of the pseudo-NMR image include: · High resolution, with a resolution of no less than 512×512 pixels; · High contrast, the contrast is not less than 256 gray levels; · It can clearly display the soft tissue and bony structure of the temporomandibular joint.

8. A method for generating pseudo nuclear magnetic resonance images based on artificial intelligence, characterized in that: The following steps are involved: · Preprocess CBCT image data by normalization, noise reduction and format conversion; · Use deep learning models to generate pseudo images with MRI characteristics; · Combined with the characteristics of the artifact images, the generated results are evaluated and adjusted through the quality optimization algorithm.

9. The method according to claim 8, characterized in that The process of generating a pseudo image by the deep learning model includes: · Using a diffusion model, images are generated through iterative denoising; · Use conditional generative adversarial networks (cGAN) to generate fake images based on input images and conditions; · The detail and clarity of the generated image are optimized in combination with boundary conditions.

10. The method according to claim 8, characterized in that The quality optimization includes comprehensively evaluating and adjusting the pseudo-images using perceptual loss, structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR); and / or The pseudo-NMR image generation process supports the deployment of a cloud computing platform to improve processing efficiency and achieve remote optimization and updating of model parameters.

Images