Medical image reconstruction method and system for radiology department

Abstract

The invention relates to the technical field of radiology department, in particular to a medical image reconstruction method and system for radiology department, and the method comprises the following steps: S1, carrying out the collection and preprocessing of multi-modal data; s2, image initial reconstruction based on sparse representation and compressed sensing; s3, deep learning is carried out to enhance the image quality; s4, real-time reconstruction and image post-processing are carried out; through the multi-modal sensor fusion and deep learning enhancement technology, the resolution, contrast and definition of medical images are remarkably improved, artifacts and noise are reduced, more accurate and clearer diagnosis basis is provided for doctors, early-stage minimal lesions can be found, diagnosis accuracy is improved, and the diagnosis efficiency is improved. A low-dose X-ray detector and ultrasonic sensor fusion technology is introduced into CT imaging, so that the radiation dose of X-rays can be effectively reduced on the premise of ensuring the image quality, radiation injury to a patient due to frequent examination is reduced, and the medical safety is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of radiology, and particularly to a medical image reconstruction method and system for radiology. Background Art

[0002] In the daily work of radiology, medical images play a crucial role in the diagnosis of diseases. Traditional medical image acquisition technologies, such as X-ray, CT (Computed Tomography), MRI (Magnetic Resonance Imaging), etc., although they can obtain images of the internal structure of the human body, have many problems. For example, X-ray images have low resolution and limited ability to display soft tissues, making it difficult to clearly present subtle lesions; while CT imaging improves resolution, it has a high risk of radiation dose, and frequent examinations may cause potential harm to patients; MRI imaging has no radiation and good imaging effect on soft tissues, but the imaging time is long, and artifacts are easily generated due to the minute movement of patients, affecting the image quality and diagnostic accuracy. In addition, existing image reconstruction methods often rely on complex algorithms and a large amount of computing resources. In actual clinical applications, it is difficult to meet the dual requirements of real-time and accuracy, resulting in low diagnostic efficiency and possible delay of the disease. Therefore, a medical image reconstruction method and system for radiology are proposed. Summary of the Invention

[0003] In view of this, the present invention provides a medical image reconstruction method and system for radiology to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0004] The technical solution of the present invention is realized as follows: A medical image reconstruction method for radiology includes the following steps:

[0005] S1. Multi-modal data acquisition and preprocessing;

[0006] S2. Initial image reconstruction based on sparse representation and compressive sensing;

[0007] S3. Deep learning to enhance image quality;

[0008] S4. Real-time reconstruction and post-image processing.

[0009] Further preferably, in the S1, a multi-modal sensor fusion technology is adopted to collect data on the same part of the human body at the same time, so that the acquired data has consistency in time and space. The collected raw data adopts an adaptive filtering algorithm to remove random noise, power frequency interference, etc. in the data, improve the signal-to-noise ratio of the data, and provide a clean and accurate data basis for subsequent image reconstruction.

[0010] Further preferably, in S2, the preprocessed data is sparsely represented, and a suitable sparse basis is selected, which can effectively compress the data while retaining the key features of the image. Based on the compressive sensing theory, on the premise of ensuring accurate image reconstruction, the sparsely represented data is sampled at a rate much lower than the traditional Nyquist sampling rate. Through a carefully designed measurement matrix, a linear projection is performed on the sparse coefficients to obtain a small number of measurement values. The measurement values contain the key information of the original image. Subsequently, by solving a series of optimization problems, the sparse representation of the original image is accurately recovered from the small number of measurement values, and then the preliminary medical image is reconstructed using the sparse representation coefficients.

[0011] Further preferably, in S3, a convolutional neural network is used as the core model for image enhancement. A large number of labeled high-quality medical images are collected as training samples. During the training process, a loss function is used to measure the difference between the model output and the true label, and the parameters of the model are continuously adjusted through the backpropagation algorithm to minimize the value of the loss function. After training with a large number of samples, the model learns the mapping relationship from low-quality images to high-quality images. In practical applications, the preliminary image reconstructed based on sparse representation and compressive sensing is input into the trained CNN model, and the model outputs the enhanced medical image, significantly improving the contrast and clarity of the image, effectively repairing the artifacts and noise in the image, and enhancing the visual quality and diagnostic usability of the image.

[0012] Further preferably, in S4, to meet the radiologists' need for immediate diagnosis, parallel computing technology is used to accelerate the image reconstruction process. The graphics processor is used to optimize the image reconstruction algorithm. Through reasonable task division and data transmission optimization, real-time reconstruction of medical images is achieved, greatly shortening the time interval from data acquisition to image display, improving the work efficiency of the radiology department. Using a professional image display library, the image is rendered, and according to the display standards of the radiology department, parameters such as the brightness, contrast, and color of the image are adjusted to ensure that the image can accurately present the details of human tissues. Doctors can observe the image from different angles and scales through these operations to conduct a more detailed analysis of the lesion site.

[0013] A medical image reconstruction system for the radiology department includes the following modules: an image acquisition optimization module, a sparse representation and compressive sensing module, a deep learning enhancement module, and a real-time reconstruction and display module.

[0014] Further preferably, the image acquisition optimization module uses sensor synchronous triggering technology to ensure that different types of sensors collect data from the same part of the human body at the same time. An adaptive filtering algorithm is used to preprocess the collected raw data, removing noise and interference signals, providing a more accurate data basis for subsequent image reconstruction.

[0015] Further preferably, for the sparse representation and compressive sensing module, a suitable sparse basis is selected to transform the image data, represent the image as a sparse vector, and based on the optimization algorithm of compressive sensing, accurately recover the sparse representation of the original image from a small number of measurement values by solving a series of optimization problems, and then reconstruct a complete high-quality image.

[0016] Further preferably, for the deep learning enhancement module, deep learning architectures such as convolutional neural networks or generative adversarial networks are adopted, and the model is trained using a large number of labeled high-quality medical image samples, enabling the model to learn the mapping relationship from low-quality images to high-quality images. In practical applications, the preliminarily reconstructed image is input into the trained model to output the enhanced image.

[0017] Further preferably, for the real-time reconstruction and display module, parallel computing technology is adopted to optimize the image reconstruction algorithm, improve the computing efficiency, achieve real-time reconstruction, and use a professional image display library to render and display the reconstructed image according to the display standards of the radiology department to ensure accurate presentation of parameters such as the color and contrast of the image.

[0018] Due to the adoption of the above technical solutions in the embodiments of the present invention, it has the following advantages:

[0019] 1. Through multi-modal sensor fusion and deep learning enhancement technologies, the present invention significantly improves the resolution, contrast, and clarity of medical images, reduces artifacts and noise, provides more accurate and clear diagnostic basis for doctors, helps detect early tiny lesions, improves diagnostic accuracy, and introduces the fusion technology of low-dose X-ray detectors and ultrasonic sensors in CT imaging. On the premise of ensuring image quality, it can effectively reduce the radiation dose of X-rays and reduce the radiation damage suffered by patients due to frequent examinations, improving medical safety.

[0020] 2. The present invention combines sparse representation and compressive sensing as well as parallel computing technology, reduces the data acquisition volume and computing time, realizes the rapid reconstruction and real-time display of medical images, improves the work efficiency of the radiology department, reduces the waiting time of patients, and is especially suitable for scenarios with high time requirements such as emergency. The image interaction operation function provided by the real-time reconstruction and display module facilitates doctors to observe and analyze the images from multiple angles and levels, improves the convenience and accuracy of diagnosis, and helps doctors make more scientific and reasonable diagnostic decisions.

[0021] The above summary is only for the purpose of the specification and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the present invention will be readily apparent by reference to the drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0023] Figure 1 It is a system module diagram of the present invention;

[0024] Figure 2 It is a method flow chart of the present invention. Detailed implementation manners

[0025] In the following, only some exemplary embodiments are briefly described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present invention. Therefore, the drawings and the description are considered to be exemplary in nature rather than restrictive.

[0026] The embodiments of the present invention will be described in detail below with reference to the drawings.

[0027] As Figure 1-2 shown, the embodiments of the present invention provide a medical image reconstruction method for the radiology department, including the following steps:

[0028] S1. Multi-modal data acquisition and preprocessing;

[0029] S2. Initial image reconstruction based on sparse representation and compressive sensing;

[0030] S3. Deep learning to enhance image quality;

[0031] S4. Real-time reconstruction and post-image processing.

[0032] In one embodiment, in S1, the multi-modal sensor fusion technology is adopted to collect data on the same part of the human body at the same time, so that the acquired data has consistency in time and space. The collected raw data uses an adaptive filtering algorithm to remove random noise, power frequency interference, etc. in the data, improve the signal-to-noise ratio of the data, and provide a clean and accurate data basis for subsequent image reconstruction.

[0033] In one embodiment, in S2, the preprocessed data is sparsely represented, and a suitable sparse basis is selected, which can effectively compress the data while retaining the key features of the image. Based on the compressive sensing theory, on the premise of ensuring accurate image reconstruction, the sparsely represented data is sampled at a rate much lower than the traditional Nyquist sampling rate. Through a carefully designed measurement matrix, a linear projection is performed on the sparse coefficients to obtain a small number of measurement values. These measurement values contain the key information of the original image. Subsequently, by solving a series of optimization problems, the sparse representation of the original image is accurately recovered from the small number of measurement values, and then the sparse representation coefficients are used to reconstruct a preliminary medical image.

[0034] In one embodiment, in S3, a convolutional neural network is used as the core model for image enhancement. A large number of labeled high-quality medical images are collected as training samples. During the training process, a loss function is used to measure the difference between the model output and the true label, and the parameters of the model are continuously adjusted through the backpropagation algorithm to minimize the value of the loss function. After training with a large number of samples, the model learns the mapping relationship from low-quality images to high-quality images. In practical applications, the preliminary image obtained by sparse representation and compressive sensing reconstruction is input into the trained CNN model, and the model outputs an enhanced medical image, significantly improving the contrast and clarity of the image, effectively repairing artifacts and noise in the image, and enhancing the visual quality and diagnostic usability of the image.

[0035] In one embodiment, in S4, in order to meet the radiologist's need for immediate diagnosis, parallel computing technology is used to accelerate the image reconstruction process. The graphics processor is used to optimize the image reconstruction algorithm. Through reasonable task partitioning and data transfer optimization, real-time reconstruction of medical images is achieved, greatly shortening the time interval from data acquisition to image display, improving the work efficiency of the radiology department. Using a professional image display library, the image is rendered, and parameters such as brightness, contrast, and color of the image are adjusted according to the display standards of the radiology department to ensure that the image can accurately present the details of human tissues. Doctors can observe the image from different angles and scales through these operations to conduct a more detailed analysis of the lesion site.

[0036] A medical image reconstruction system for the radiology department includes the following modules: an image acquisition optimization module, a sparse representation and compressive sensing module, a deep learning enhancement module, and a real-time reconstruction and display module.

[0037] In one embodiment, the image acquisition optimization module uses sensor synchronous triggering technology to ensure that different types of sensors collect data from the same part of the human body at the same time. An adaptive filtering algorithm is used to preprocess the collected raw data to remove noise and interference signals, providing a more accurate data basis for subsequent image reconstruction.

[0038] In one embodiment, the sparse representation and compressive sensing module selects an appropriate sparse basis to transform the image data, represents the image as a sparse vector, and based on the optimization algorithm of compressive sensing, accurately recovers the sparse representation of the original image from a small number of measurement values by solving a series of optimization problems, and then reconstructs a complete high-quality image.

[0039] In one embodiment, the deep learning enhancement module adopts deep learning architectures such as convolutional neural networks or generative adversarial networks, and uses a large number of labeled high-quality medical image samples to train the model, enabling the model to learn the mapping relationship from low-quality images to high-quality images. In practical applications, the preliminarily reconstructed image is input into the trained model, and the enhanced image is output.

[0040] In one embodiment, the real-time reconstruction and display module adopts parallel computing technology to optimize the image reconstruction algorithm, improve the computing efficiency, achieve real-time reconstruction, and uses a professional image display library to render and display the reconstructed image according to the display standards of the radiology department to ensure the accurate presentation of parameters such as the color and contrast of the image.

[0041] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various changes or substitutions, and these should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. a. A medical image reconstruction method for the radiology department, characterized in that: Including the following steps: S1. Multimodal data acquisition and preprocessing; S2. Initial image reconstruction based on sparse representation and compressive sensing; S3. Deep learning for enhancing image quality; S4. Real-time reconstruction and post-processing of images.

2. The medical image reconstruction method for the radiology department according to claim 1, characterized in that: In the above S1, the multimodal sensor fusion technology is adopted to collect data on the same part of the human body at the same moment, so that the acquired data has consistency in time and space. The collected raw data uses an adaptive filtering algorithm to remove random noise, power frequency interference, etc. in the data, improve the signal-to-noise ratio of the data, and provide a clean and accurate data basis for subsequent image reconstruction.

3. A medical image reconstruction method for a radiology department according to claim 1, characterized in that: In the above S2, the preprocessed data is sparsely represented, and a suitable sparse basis is selected, which can effectively compress the data while retaining the key features of the image. Based on the compressive sensing theory, on the premise of ensuring accurate image reconstruction, the sparsely represented data is sampled at a rate much lower than the traditional Nyquist sampling rate. Through a carefully designed measurement matrix, a linear projection is performed on the sparse coefficients to obtain a small number of measurement values. The measurement values contain the key information of the original image. Subsequently, by solving a series of optimization problems, the sparse representation of the original image is accurately recovered from the small number of measurement values, and then the initial medical image is reconstructed using the sparse representation coefficients.

4. A medical image reconstruction method for a radiology department according to claim 1, characterized in that: In the above S3, a convolutional neural network is used as the core model for image enhancement. A large number of labeled high-quality medical images are collected as training samples. During the training process, a loss function is used to measure the difference between the model output and the true label, and the parameters of the model are continuously adjusted through the backpropagation algorithm to minimize the value of the loss function. After training with a large number of samples, the model learns the mapping relationship from low-quality images to high-quality images. In practical applications, the initial image reconstructed based on sparse representation and compressive sensing is input into the trained CNN model, and the model outputs the enhanced medical image, significantly improving the contrast and clarity of the image, effectively repairing artifacts and noise in the image, and enhancing the visual quality and diagnostic usability of the image.

5. A medical image reconstruction method for the radiology department according to claim 1, characterized in that: In the above S4, in order to meet the needs of radiologists for immediate diagnosis, parallel computing technology is adopted to accelerate the image reconstruction process. The graphics processor is used to optimize the image reconstruction algorithm. Through reasonable task division and data transmission optimization, real-time reconstruction of medical images is achieved, greatly shortening the time interval from data acquisition to image display, improving the work efficiency of the radiology department. Using a professional image display library, the image is rendered, and according to the display standards of the radiology department, parameters such as the brightness, contrast, and color of the image are adjusted to ensure that the image can accurately present the details of human tissues. Doctors can observe the image from different angles and scales through these operations to conduct a more detailed analysis of the lesion site.

6. A medical image reconstruction system for the radiology department, which is used in conjunction with a medical image reconstruction method for the radiology department according to any one of claims 1-5, characterized in that: Including the following modules: Image acquisition optimization module, sparse representation and compressive sensing module, deep learning enhancement module, and real-time reconstruction and display module.

7. A medical image reconstruction system for a radiology department according to claim 6, characterized in that: The image acquisition optimization module uses sensor synchronization triggering technology to ensure that different types of sensors collect data from the same part of the human body at the same moment. It adopts an adaptive filtering algorithm to preprocess the collected raw data, removing noise and interference signals, and providing a more accurate data basis for subsequent image reconstruction.

8. The medical image reconstruction system for the radiology department according to claim 6, characterized in that: The sparse representation and compressive sensing module selects an appropriate sparse basis to transform the image data, representing the image as a sparse vector. Based on the optimization algorithm of compressive sensing, by solving a series of optimization problems, it accurately recovers the sparse representation of the original image from a small number of measurement values, and then reconstructs a complete high-quality image.

9. A medical image reconstruction system for a radiology department according to claim 6, characterized in that: The deep learning enhancement module adopts deep learning architectures such as convolutional neural networks or generative adversarial networks, and uses a large number of labeled high-quality medical image samples to train the model, enabling the model to learn the mapping relationship from low-quality images to high-quality images. In practical applications, the preliminarily reconstructed image is input into the trained model, and the enhanced image is output.

10. A medical image reconstruction system for the radiology department according to claim 6, characterized in that: The real-time reconstruction and display module uses parallel computing technology to optimize the image reconstruction algorithm, improve the computing efficiency, and achieve real-time reconstruction. Using a professional image display library, it renders and displays the reconstructed image according to the display standards of the radiology department, ensuring accurate presentation of parameters such as the color and contrast of the image.

Images