A medical image processing method and system

Abstract

A medical image processing method and system, the method comprising: acquiring an original medical image; performing optimization processing on the original medical image to obtain an output image of the original medical image; inputting the output image and output K-space data of the output image into a fidelity model to obtain a fidelity image of the output image, wherein the fidelity model is trained based on sample images and sample K-space data.

Description

Technical Field

[0001] This manual relates to the field of medical technology, and in particular to a medical image processing method and system. Background Technology

[0002] Deep learning methods have been applied to denoising, artifact removal, and super-resolution of medical images. However, deep neural methods utilize direct mapping from input to output, raising concerns about their reliability. Furthermore, the processed medical images may exhibit distortion.

[0003] Therefore, there is an urgent need for an image processing method to improve image quality. Summary of the Invention

[0004] One embodiment of this specification provides a medical image processing method, the method comprising: acquiring an original medical image; optimizing the original medical image to obtain an output image of the original medical image; inputting the output image and the output K-space data of the output image into a fidelity model to obtain a fidelity image of the output image, wherein the fidelity model is trained based on sample images and sample K-space data.

[0005] One embodiment of this specification provides a medical image processing system, comprising: an acquisition module for acquiring an original medical image; optimizing the original medical image to obtain an output image of the original medical image; and a fidelity module for inputting the output image and its output K-space data into a fidelity model to obtain a fidelity image of the output image, wherein the fidelity model is trained based on sample images and sample K-space data.

[0006] One embodiment of this specification provides a medical image processing apparatus, characterized in that the apparatus includes: at least one storage medium storing computer instructions; and at least one processor executing the computer instructions to implement the medical image processing method as described above.

[0007] One embodiment of this specification provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions, the computer executes the medical image processing method as described above.

[0008] In existing technologies, the main method for ensuring fidelity is to use constraints and backfilling of medical image information in the network.

[0009] This specification presents a medical image processing method and system that can perform fidelity processing on medical images after conventional processing (denoising, artifact removal, super-resolution, etc.) to achieve fidelity preservation of medical images and improve image processing results. Attached Figure Description

[0010] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:

[0011] Figure 1 These are schematic diagrams illustrating application scenarios of exemplary medical image processing systems according to some embodiments of this specification;

[0012] Figure 2 This is a block diagram of an exemplary medical image processing system according to some embodiments of this specification;

[0013] Figure 3 This is a flowchart illustrating an exemplary medical image processing method according to some embodiments of this specification;

[0014] Figure 4 This is a schematic diagram of a medical image processing method according to some embodiments of this specification;

[0015] Figure 5 This is a schematic diagram illustrating the training of a fidelity model according to some embodiments of this specification. Detailed Implementation

[0016] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0017] It should be understood that the terms “system,” “device,” “unit,” and / or “module” used herein are one way to distinguish different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.

[0018] As indicated in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0019] Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0020] Figure 1 This is a schematic diagram illustrating an application scenario of an exemplary medical image processing system according to some embodiments of this specification. In some embodiments, such as... Figure 1 As shown, the application scenario 100 of the medical image processing system may include at least a scanning device 110, a processing device 120, a terminal device 130, a storage device 140, and a network 150.

[0021] The scanning device 110 can scan a target object within a detection area or a scanning area to obtain scan data of the target object. In some embodiments, the target object may include biological objects and / or non-biological objects. For example, the target object may be living or non-living organic and / or inorganic matter.

[0022] In some embodiments, the scanning device 110 may include any combination thereof, such as a permanent magnet magnetic resonance imaging (MRI) device, a conventional magnetic resonance imaging (MRI) device, a superconducting magnetic resonance imaging (MRI) device, a nuclear magnetic resonance imaging (NMR) device, an electron paramagnetic resonance imaging (EPI) device, or an electron spin resonance imaging (ESI) device. In some embodiments, the scanning device 110 may include a single-modal scanner and / or a multimodal scanner. A single-modal scanner may include, for example, an MRI scanner. A multimodal scanner may include, for example, an X-ray imaging-MRI scanner, a single-photon emission computed tomography-MRI scanner, a digital subtraction angiography-MRI scanner, or any combination thereof. The above description of the scanning devices is for illustrative purposes only and is not intended to limit the scope of this specification.

[0023] The processing device 120 can process data and / or information acquired from the scanning device 110, the terminal device 130, the storage device 140, and / or other components of the application scenario 100 of the medical image processing system. For example, the processing device 120 can acquire images (e.g., raw medical images, output images, fused images, gold standard images, etc.) from the terminal device 130 and the storage device 140, and analyze and process them.

[0024] In some embodiments, processing device 120 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, processing device 120 may be local or remote. For example, processing device 120 may access information and / or data from scanning device 110, terminal device 130, and / or storage device 140 via network 150. Alternatively, processing device 120 may be directly connected to scanning device 110, terminal device 130, and / or storage device 140 to access information and / or data. In some embodiments, processing device 120 may be implemented on a cloud platform. For example, the cloud platform may include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, inter-cloud cloud, multi-cloud, etc., or any combination thereof.

[0025] In some embodiments, the processing device 120 and the scanning device 110 may be integrated into one unit. In some embodiments, the processing device 120 and the scanning device 110 may be directly or indirectly connected to work together to implement the methods and / or functions described herein.

[0026] In some embodiments, the processing device 120 may include input devices and / or output devices. These input devices and / or output devices enable interaction with the user (e.g., setting image fusion levels, labeling gold standard images, etc.). In some embodiments, the input devices and / or output devices may include a display screen, keyboard, mouse, microphone, etc., or any combination thereof.

[0027] Terminal device 130 can communicate and / or connect to scanning device 110, processing device 120, and / or storage device 140. In some embodiments, interaction with a user can be achieved through terminal device 130. In some embodiments, terminal device 130 may include mobile device 131, tablet computer 132, laptop computer 133, etc., or any combination thereof. In some embodiments, terminal device 130 (or all or part of its functions) may be integrated into processing device 120.

[0028] Storage device 140 may store data, instructions, and / or any other information. In some embodiments, storage device 140 may store data acquired from scanning device 110, processing device 120, terminal device 130, and / or other sources (e.g., raw medical images, output images, fused images, gold standard images, k-space data, etc.). In some embodiments, storage device 140 may store data and / or instructions used by processing device 120 to perform or use in order to complete the exemplary methods described herein.

[0029] In some embodiments, storage device 140 may include one or more storage components, each of which may be a separate device or part of another device. In some embodiments, storage device 140 may include random access memory (RAM), read-only memory (ROM), mass storage, removable memory, volatile read-write memory, and any combination thereof. In some embodiments, storage device 140 may be implemented on a cloud platform. In some embodiments, storage device 140 may be part of scanning device 110, processing device 120, and / or terminal device 130.

[0030] Network 150 may include any suitable network capable of facilitating information and / or data exchange. In some embodiments, at least one component of application scenario 100 of the medical image processing system (e.g., scanning device 110, processing device 120, terminal device 130, storage device 140) may exchange information and / or data with at least one other component of application scenario 100 of the medical image processing system via network 150. For example, processing device 120 may acquire raw medical images from scanning device 110 via network 150.

[0031] It should be noted that the above description of application scenario 100 of the medical image processing system is provided for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can make various modifications or variations based on the description in this specification. For example, application scenario 100 of the medical image processing system can achieve similar or different functions on other devices. However, these changes and modifications will not depart from the scope of this specification.

[0032] Figure 2 This is a block diagram of an exemplary medical image processing system according to some embodiments of this specification. For example... Figure 2 As shown, in some embodiments, the medical image processing system 200 may include an acquisition module 210 and a fidelity module 220. In some embodiments, the functions corresponding to the medical image processing system 200 may be performed by the processing device 120.

[0033] The acquisition module 210 can be used to acquire raw medical images and optimize them to obtain an output image. For more information on acquiring raw medical images, please refer to [link to relevant documentation]. Figure 3 Step 310 and its related description. For more information on optimizing raw medical images, please refer to [link / reference needed]. Figure 3 Step 320 and its related description.

[0034] The fidelity module 220 can be used to input the output image and its output K-space data into the fidelity model to obtain a fidelity image of the output image. The fidelity model is trained based on sample images and sample K-space data. For more information on fidelity image acquisition, please refer to [link to relevant documentation]. Figure 3 Step 330 and its related description.

[0035] It should be understood that Figure 2 The systems and modules shown can be implemented in various ways. For example, they can be implemented by hardware, software, or a combination of both. The systems and modules in this specification can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips and transistors, or programmable hardware devices such as field-programmable gate arrays and programmable logic devices, but also by software, for example, executed by various types of processors, or by a combination of the aforementioned hardware circuits and software (e.g., firmware).

[0036] It should be noted that the above description of the system and its modules is for illustrative purposes only and should not be construed as limiting this specification to the scope of the illustrated embodiments. It is understood that those skilled in the art, after understanding the principles of this system, may arbitrarily combine the various modules or construct subsystems connected to other modules without departing from these principles.

[0037] Figure 3 This is a flowchart illustrating an exemplary medical image processing method according to some embodiments of this specification. In some embodiments, process 300 may be executed by processing device 120 or medical image processing system 200. For example, process 300 may be stored in a storage device (e.g., storage device 140, storage unit of processing device 120) in the form of a program or instructions, and executed by a processor or... Figure 2 When the module shown executes a program or instructions, it can implement process 300. In some embodiments, process 300 may be completed using one or more additional operations not described below, and / or not through one or more operations discussed below. Additionally, as Figure 3 The order of operations shown is not restrictive.

[0038] Step 310: Acquire the raw medical image. In some embodiments, step 310 may be performed by the processing device 120 or the acquisition module 210.

[0039] Raw medical images refer to medical images that require processing (denoising, artifact removal, and super-resolution). For example, raw medical images can be images reconstructed from image data acquired by scanning equipment 110, or images sent from lower-level medical institutions to higher-level medical institutions for optimization. The types of raw medical images can include MRI images, X-ray-MRI images, SPECT-MRI images, DSA-MRI images, etc. Due to patient movement, foreign bodies inside or outside the body, and / or equipment limitations, raw medical images may contain artifacts and noise; the signal-to-noise ratio and / or resolution of the raw medical images may not meet standards.

[0040] In some embodiments, the original medical image can be obtained from the storage unit of the scanning device 110, the storage device 140, or the processing device 120. In some embodiments, the acquisition module 210 can obtain the original medical image by reading from the storage device or the database, or by calling a data interface.

[0041] Step 320: Optimize the original medical image to obtain an output image of the original medical image. In some embodiments, step 320 may be performed by the processing device 120 or the acquisition module 210.

[0042] Optimization processing can include denoising, super-resolution, and other techniques. The output image obtained after optimization processing has better quality than the original medical image, for example, a higher signal-to-noise ratio, higher resolution, fewer artifacts, and less noise.

[0043] In some embodiments, the optimization process can be implemented using machine learning. In other words, the optimization process can be implemented based on a trained machine learning model. The machine learning model can be, but is not limited to, one or more combinations of neural network models, support vector machine models, k-nearest neighbor models, decision tree models, etc. Deep neural networks can include one or more combinations of CNN, LeNet, GoogLeNeT, ImageNet, AlexNet, VGG, ResNet, RNN, GAN, etc.

[0044] Step 330: Input the output image and its output K-space data into the fidelity model to obtain a fidelity image of the output image, wherein the fidelity model is trained based on the sample image and the sample K-space data. In some embodiments, step 330 may be performed by the processing device 120 or the fidelity module 220.

[0045] In some embodiments, the output K-space data of the output image can be obtained by performing a Fourier transform on the output image.

[0046] A fidelity model refers to a neural network model used for medical image fidelity preservation. Deep neural networks can include one or more combinations of CNN, LeNet, GoogLeNeT, ImageNet, AlexNet, VGG, ResNet, RNN, GAN, etc.

[0047] In some embodiments, the fidelity model may have the same structure as the deep neural network used to optimize the original medical image, and the optimization process may include denoising, artifact removal, super-resolution, and other processing. In some embodiments, the fidelity model may be a post-processing step of the optimization process; for example, the original medical image may be optimized and then subjected to fidelity processing by the fidelity model.

[0048] In some embodiments, the loss function of the fidelity model includes a loss function in the image domain and a loss function in the data domain. In some embodiments, the fidelity module 220 can add or weightedly add the loss function in the image domain and the loss function in the data domain to obtain the loss function of the fidelity model. For example, as... Figure 4 As shown, the loss function 410 in the image domain and the loss function 420 in the data domain are added together or weighted together to obtain the loss function of the fidelity model.

[0049] In some embodiments, the image domain loss function is obtained based on gold-standard image samples and output image samples. For example, such as Figure 4 As shown, the image domain loss function 410 can be obtained based on the gold standard image sample 411 and the output image sample 412. For example, the fidelity module can perform mean square error, cross entropy, and other processing on the gold standard image sample 411 and the output image sample 412 to obtain the image domain loss function 410.

[0050] A gold standard image sample refers to an image sample used for reference. Compared to the original medical image, the gold standard image has a higher signal-to-noise ratio, higher resolution, etc. In some embodiments, the gold standard image sample can be obtained by optimizing the original medical image sample (e.g., denoising, artifact removal, super-resolution). In some embodiments, the above optimization process can be performed manually, for example, by a doctor, medical imaging expert, etc.

[0051] In some embodiments, the loss function for the data domain is obtained based on the output K-space data samples and the label K-space data samples. For example, as... Figure 4 As shown, the loss function 420 in the data domain can be obtained based on the output K-space data sample 421 and the label K-space data sample 422. For example, the fidelity module can perform mean squared error, cross-entropy, and other processing on the output K-space data sample 421 and the label K-space data sample 422 to obtain the loss function 420 in the data domain.

[0052] For training methods of the high-fidelity model, please refer to [link / reference]. Figure 5 And its explanation.

[0053] It should be noted that the above description of process 300 is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to process 300 under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0054] Figure 5 This is a schematic diagram illustrating the training of a fidelity model according to some embodiments of this specification.

[0055] like Figure 5 As shown, in some embodiments, an initial fidelity model 510 can be trained based on a large number of labeled training samples to update the parameters of the initial fidelity model and obtain a trained fidelity model 520.

[0056] In some embodiments, the processing device may acquire multiple training samples, each training sample including a sample image and sample K-space data; the sample image includes an output image sample after optimization processing of the original medical image sample, and a gold standard image sample corresponding to the output image sample; the sample K-space data includes an output K-space data sample corresponding to the output image sample, and a label K-space data sample; the label K-space data sample is determined based on the original medical image sample and the output image sample.

[0057] In some embodiments, the processing device may determine the label K-space data sample based on the original K-space data sample corresponding to the original medical image sample and the output K-space data sample corresponding to the output image sample.

[0058] In some embodiments, the processing device 120 may use the following two methods or any combination thereof to obtain tag K-space data samples.

[0059] Method 1: First, the processing device 120 can acquire the original K-space data sample corresponding to the original medical image sample and the output K-space data sample corresponding to the output image sample.

[0060] In some embodiments, the original K-space data sample can be obtained by performing a Fourier transform on the original medical image sample. In some embodiments, the original K-space data sample can be obtained by scanning device 110 and / or processing device 120, for example, scanning device 110 and / or processing device 120 can output the intermediate result (i.e., K-space data without inverse Fourier transform) of the original medical image sample acquisition process.

[0061] In some embodiments, the output K-space data samples of the output image samples can be obtained by performing a Fourier transform on the output image samples.

[0062] Secondly, the processing device 120 can fuse the original K-space data sample and the output K-space data sample to obtain a tag K-space data sample. Different fusion ratios can be used to obtain the tag K-space data sample according to application requirements. In one embodiment, the fusion can be divided into seven levels according to different ratios of the original K-space data sample and the output K-space data sample. For example, there are six levels: Level 0 (output K-space data sample: original K-space data sample = 0%:100%); Level 1 (output K-space data sample: original K-space data sample = 10%:90%); Level 2 (output K-space data sample: original K-space data sample = 30%:70%); Level 3 (output K-space data sample: original K-space data sample = 50%:50%); Level 4 (output K-space data sample: original K-space data sample = 70%:30%); Level 5 (output K-space data sample: original K-space data sample = 90%:10%); and Level 6 (output K-space data sample: original K-space data sample = 100%:0%). The Level 0 image is equivalent to the original image sample, Level 1 is closer to the original image sample, Level 2 is next, Levels 3 and 4 gradually approach the output image sample, Level 5 is closest to the output image sample, and Level 6 is equivalent to the output image sample. In some embodiments, the gear level is an open parameter that can be set according to needs and / or the doctor's habits and preferences. In some embodiments, the initial default gear level is 2.

[0063] In one embodiment, the K-space is divided into at least two regions, and a specific proportional relationship is set for each region. The original K-space data sample and the output K-space data sample are then fused to obtain a label K-space data sample. For example, the K-space is divided into a first region and a second region. The K-space data of the first region is fused using a first proportional relationship (e.g., level 1), and the K-space data of the second region is fused using a second proportional relationship (e.g., level 5), thereby obtaining a label K-space data sample. The first and second proportional relationships can be set according to the characteristics of the K-space data. Alternatively, only a portion of the K-space data can be fused, similar to selecting the aforementioned level 0 or level 6 proportional relationship when fusion of a portion of at least two regions.

[0064] In some embodiments, the K-space is divided into a central region and a peripheral region, and K-space data fusion is performed on the central region. For example, data fusion is performed on the central region using a zero-level setting, resulting in the K-space central data of the labeled K-space data sample being the same as the K-space central data of the original medical image sample. K-space central data refers to the K-space data in the central part of the K-space, such as data within a certain distance from the K-space center point. Since the K-space center determines the overall structure and contrast of a medical image (e.g., an MRI image), replacing the K-space central data of the output image with the K-space central data of the original medical image can largely preserve the overall structure and contrast information of the original image, improving the fidelity of the neural network output image.

[0065] In some embodiments, the K-space peripheral region may not be fused, or data fusion may be performed using a different proportional relationship than that of the K-space central region. Since the K-space peripheral region is related to image details, in order to improve image processing effects (e.g., denoising, artifact removal, super-resolution, etc.), it is preferable to set the weight of the K-space peripheral region of the output image to be greater than the weight of the K-space peripheral region of the original image for weighted data fusion, thereby preserving the image optimization processing effect to a greater extent.

[0066] Method 2: First, the processing device 120 can fuse the original medical image samples and the output image samples based on various fusion methods to obtain a fused image sample. In some embodiments, the fusion method may include principal component transform fusion, product transform fusion, wavelet transform fusion, Laplace transform fusion, or any combination thereof. For example, the original medical image samples and the output image samples can be fused using the principal component transform fusion method to obtain a fused image sample.

[0067] Secondly, the processing device 120 can obtain the fused K-space data of the corresponding fused image sample, i.e., the label K-space data sample, by performing Fourier transform on the fused image sample.

[0068] In some embodiments, the processing device can input multiple training samples into an initial fidelity model for iterative training. Once preset conditions are met, a trained fidelity model is obtained. Preset conditions may include loss function convergence or loss function being less than a preset threshold.

[0069] In some embodiments, the processing device may obtain multiple training samples, including their corresponding labels (e.g., gold standard images of samples, label K-space data samples), by reading from a database or storage device or calling a data interface.

[0070] In some embodiments, the processing device can optimize the original medical image sample 511 to obtain the output image sample 512. Alternatively, in some embodiments, the processing device can input the original medical image 511 into a model for image optimization processing and an initial fidelity model to obtain the output image sample 512.

[0071] In some embodiments, the processing device may construct an image domain loss function and a data domain loss function based on the labels of the output image sample 512 and the training sample, and update the initial fidelity model according to the image domain loss function and the data domain loss function to obtain the fidelity model.

[0072] In some embodiments, a fidelity model can also be obtained according to other training methods, such as setting an initial learning rate (e.g., 0.1) and a learning rate decay strategy for the fidelity model, and obtaining the fidelity model by joint training with a model used for optimization based on labeled training samples. This application does not impose any limitations herein.

[0073] It should be noted that the above description of process 500 is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to process 500 under the guidance of this specification. However, these modifications and changes are still within the scope of this specification. In some embodiments, Figure 5 The model generation process described in the text is related to image processing (e.g., Figure 3 The generation of the output image (generating a high-fidelity image) described herein can be performed on different processing devices. For example, Figure 5 The model generation process described herein can be performed on the processing equipment of the scanning equipment manufacturer, while part or all of the image processing can be performed on the processing equipment of the scanning equipment user (e.g., a hospital).

[0074] In some embodiments of this specification, (1) using the K-space center data of the original medical image as a constraint of the deep neural network is beneficial to achieving image fidelity at the level of overall structure and contrast; (2) updating the fidelity model based on the image domain loss function and the data domain loss function makes the fidelity effect more natural; (3) the fusion level can be set according to user needs and / or preferences, and users can choose the fidelity effect to be closer to the original image or closer to the output image, which can more flexibly meet user needs.

[0075] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0076] Furthermore, this specification uses specific terms to describe embodiments thereof. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Moreover, certain features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

[0077] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0078] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments in this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0079] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used in the description of embodiments are modified in some examples with the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of range in some embodiments of this specification are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

[0080] For each patent, patent application, patent application publication, and other material, such as articles, books, specifications, publications, and documents, referenced in this specification, the entire contents of which are incorporated herein by reference. This excludes historical application documents that are inconsistent with or conflict with the content of this specification, as well as documents that limit the broadest scope of the claims in this specification (currently or subsequently appended to this specification). It should be noted that in the event of any inconsistency or conflict between the descriptions, definitions, and / or terminology used in the supplementary materials to this specification and the content of this specification, the descriptions, definitions, and / or terminology used in this specification shall prevail.

[0081] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A medical image processing method, characterized in that, The method includes: Acquire raw medical images; The original medical image is optimized to obtain the output image of the original medical image; The output image and its output K-space data are input into the trained fidelity model to obtain a fidelity image of the output image. The fidelity model is a neural network model, which is trained based on sample images and sample K-space data. The sample images include original medical image samples, optimized output image samples, and gold standard image samples corresponding to the output image samples. The sample K-space data includes output K-space data samples corresponding to the output image samples and label K-space data samples. The K-space center data of the label K-space data samples is the K-space center data of the original medical image samples.

2. The medical image processing method as described in claim 1, characterized in that, The training process of the fidelity model includes: Multiple training samples are acquired, each training sample including the sample image and the sample K-space data; the sample image includes the output image sample after optimization of the original medical image sample, and the gold standard image sample corresponding to the output image sample; the sample K-space data includes the output K-space data sample corresponding to the output image sample, and the label K-space data sample; the label K-space data sample is determined based on the original medical image sample and the output image sample; The multiple training samples are input into the initial fidelity model for iterative training. When the preset conditions are met, the trained fidelity model is obtained.

3. The medical image processing method as described in claim 2, characterized in that, The K-space data samples are determined based on the original medical image samples and the output image samples, including: The label K-space data sample is determined based on the original K-space data sample corresponding to the original medical image sample and the output K-space data sample corresponding to the output image sample.

4. The medical image processing method as described in claim 2, characterized in that, The loss function of the fidelity model includes a loss function in the image domain and a loss function in the data domain. The fidelity model is trained based on the loss function in the image domain and the loss function in the data domain.

5. The medical image processing method as described in claim 4, characterized in that, The loss function in the image domain is obtained based on the output image samples and the gold standard image samples, and / or the loss function in the data domain is obtained based on the output K-space data samples and the label K-space data samples.

6. The medical image processing method as described in claim 3, characterized in that, The K-space center data of the labeled K-space data sample is the K-space center data of the original K-space data sample.

7. The medical image processing method as described in claim 1, characterized in that, The optimization of the original medical image to obtain the output image is achieved through machine learning.

8. A medical image processing system, characterized in that, The system includes: The acquisition module is used to acquire raw medical images; The original medical image is optimized to obtain the output image of the original medical image; The fidelity module is used to input the output image and the output K-space data of the output image into a trained fidelity model to obtain a fidelity image of the output image. The fidelity model is a neural network model, which is trained based on sample images and sample K-space data. The sample images include original medical image samples, optimized output image samples, and gold standard image samples corresponding to the output image samples. The sample K-space data includes output K-space data samples corresponding to the output image samples and label K-space data samples. The K-space center data of the label K-space data samples is the K-space center data of the original medical image samples.

9. A medical image processing device, characterized in that, The device includes: The display device displays at least one of the following: an original medical image, an output image of the original medical image after optimization processing, and a high-fidelity image of the output image. At least one storage medium storing computer instructions; and At least one processor executes the computer instructions to implement the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing computer instructions that, when read by a computer, execute the method as described in any one of claims 1 to 7.

Images