Magnetic resonance image optimization method, device, electronic device and storage medium

Abstract

The present invention provides a magnetic resonance image optimization method, device, electronic device and storage medium, the method comprising: obtaining a first magnetic resonance image optimization model based on a first nuclide; constructing a second magnetic resonance image optimization model based on the first magnetic resonance image optimization model based on a deep transfer learning algorithm; obtaining a magnetic resonance image to be optimized based on a second nuclide, and optimizing the magnetic resonance image to be optimized through the second magnetic resonance image optimization model; wherein the concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide. The present invention improves the optimization efficiency of the magnetic resonance image to be optimized and reduces the optimization cost.

Description

Technical Field

[0001] The present invention relates to the technical field of magnetic resonance imaging, and in particular to a magnetic resonance image optimization method, device, electronic equipment and storage medium. Background Art

[0002] At present, nuclear magnetic resonance imaging (MRI) has been widely used in medical diagnosis. Nuclear magnetic resonance imaging is a biomagnetic nuclear spin imaging technology that has developed rapidly with the development of computer technology, electronic circuit technology, and superconductor technology. It uses magnetic field and radio frequency pulses to make the nuclides in the biological tissue resonate to generate radio frequency signals, which are then processed by computer to form images. And because water is abundant in the body, and hydrogen nuclei have high nuclear magnetic sensitivity and strong signals, people prefer hydrogen nuclei ( 1 H) as the imaging nuclide for nuclear magnetic resonance imaging of living organisms. As the requirements for nuclear magnetic resonance imaging become higher and higher, the use of hydrogen nuclei as imaging nuclides cannot meet the requirements. Therefore, it is proposed to use other nuclides in the body besides hydrogen nuclei (such as sodium nuclei that are widely involved in the physiological and pathological activities of living organisms) as imaging nuclides to meet the different needs of magnetic resonance imaging.

[0003] However, since the concentration of other nuclides in the organism is low, and the transverse relaxation time is short and the gyromagnetic ratio is small, when other nuclides are subjected to magnetic resonance imaging by a magnetic resonance imaging device suitable for hydrogen nuclei, the image quality of the generated magnetic resonance images is low. There are two common ways to optimize magnetic resonance images in the prior art. One way is to increase the field strength of the magnetic resonance device and the performance of the probe to optimize the magnetic resonance images and improve the image quality of the magnetic resonance images. Another way is to optimize the magnetic resonance images based on a neural network model to improve the image quality of the magnetic resonance images.

[0004] The first method of optimizing magnetic resonance images has the following problems: it increases the cost of magnetic resonance equipment, thereby increasing the cost of optimizing magnetic resonance images. The second method of optimizing magnetic resonance images has the following problems: in order to improve the accuracy of the neural network model, a large number of magnetic resonance images generated based on other nuclides are required, and since the transverse relaxation time of other nuclides is short, the efficiency of obtaining magnetic resonance images is low, which in turn causes low efficiency in optimizing magnetic resonance images. Summary of the invention

[0005] In view of this, it is necessary to provide a magnetic resonance image optimization method, device, electronic device and storage medium to solve the technical problems of high cost and low efficiency in optimizing magnetic resonance images in the prior art.

[0006] In one aspect, the present invention provides a method for optimizing a magnetic resonance image, comprising:

[0007] acquiring a first magnetic resonance image optimization model based on a first nuclear species;

[0008] Based on the deep transfer learning algorithm, construct a second magnetic resonance image optimization model according to the first magnetic resonance image optimization model;

[0009] Acquiring a magnetic resonance image to be optimized based on a second nuclear species, and optimizing the magnetic resonance image to be optimized by using the second magnetic resonance image optimization model;

[0010] Wherein, the concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

[0011] In some possible implementations, the first nuclide is 1 H, the second nuclide is 23 Nah.

[0012] In some possible implementations, the step of constructing a second magnetic resonance image optimization model based on the first magnetic resonance image optimization model based on a deep transfer learning algorithm includes:

[0013] Acquiring a model structure and model parameters of the first magnetic resonance image optimization model;

[0014] Based on the deep transfer learning algorithm, construct a transfer model according to the model structure and the model parameters;

[0015] A second magnetic resonance image set based on the second nuclear species is acquired, and the migration model is trained by using the second magnetic resonance image set to obtain the second magnetic resonance image optimization model.

[0016] In some possible implementations, the second magnetic resonance image set includes a plurality of second magnetic resonance images; and the step of training the migration model using the second magnetic resonance image set to obtain the second magnetic resonance image optimization model includes:

[0017] selecting a plurality of magnetic resonance image samples from the plurality of second magnetic resonance images;

[0018] Inputting the magnetic resonance image samples into the migration model for training to obtain training magnetic resonance images;

[0019] determining whether training of the migration model is complete based on the training magnetic resonance images and the magnetic resonance image samples;

[0020] When the migration model has been fully trained, the fully trained migration model is the second magnetic resonance image optimization model.

[0021] In some possible implementations, the training magnetic resonance image includes a current training resonance sub-image and a previous training resonance sub-image; and determining whether the training of the migration model is complete based on the training magnetic resonance image and the magnetic resonance image sample includes:

[0022] determining a first peak signal-to-noise ratio between the last training resonance sub-image and the magnetic resonance image sample;

[0023] determining a second peak signal-to-noise ratio between the current training resonance subimage and the magnetic resonance image sample;

[0024] Determining whether a difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than a threshold difference;

[0025] If the difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than a threshold difference, the migration model has been fully trained.

[0026] In some possible implementations, after selecting a plurality of magnetic resonance image samples from the plurality of second magnetic resonance images, the method further includes:

[0027] Performing grayscale transformation on the magnetic resonance image sample to obtain a grayscale image;

[0028] Performing histogram equalization processing on the grayscale image to obtain a balanced image;

[0029] The equalized image is segmented and calibrated by an edge detection algorithm to obtain a valid image.

[0030] In some possible implementations, after the migration model has been trained, the method further includes:

[0031] Selecting a plurality of magnetic resonance test images from the plurality of second magnetic resonance images;

[0032] The model performance of the second magnetic resonance image optimization model is verified based on the magnetic resonance test image.

[0033] On the other hand, the present invention also provides a magnetic resonance image optimization device, comprising:

[0034] A first optimization model building unit, used for acquiring a first magnetic resonance image optimization model based on a first nuclear species;

[0035] A second optimization model construction unit is configured to construct a second magnetic resonance image optimization model according to the first magnetic resonance image optimization model based on a deep transfer learning algorithm;

[0036] a magnetic resonance image optimization unit, configured to obtain a magnetic resonance image to be optimized based on a second nuclear species, and optimize the magnetic resonance image to be optimized by using the second magnetic resonance image optimization model;

[0037] Wherein, the concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

[0038] On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein:

[0039] The memory is used to store programs;

[0040] The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the magnetic resonance image optimization method described in any of the above implementations.

[0041] On the other hand, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps in the magnetic resonance image optimization method described in any of the above implementations.

[0042] The beneficial effect of adopting the above embodiment is that the magnetic resonance image optimization method provided by the present invention first obtains a first magnetic resonance image optimization model based on a first nuclide with a large concentration, a long transverse relaxation time and a large gyromagnetic ratio in a biological body, and then builds a second magnetic resonance image optimization model based on the first magnetic resonance image optimization model based on a deep transfer learning algorithm, and optimizes the magnetic resonance image to be optimized generated by the second nuclide with a low concentration, a short transverse relaxation time and a small gyromagnetic ratio in a biological body through the second magnetic resonance image optimization model, without modifying the existing magnetic resonance equipment, and therefore, without increasing the cost of optimizing the magnetic resonance image to be optimized. Furthermore, since the present invention constructs the second magnetic resonance image optimization model based on the deep transfer learning algorithm, the number of magnetic resonance images based on the second nuclide required to construct the second magnetic resonance image optimization model can be greatly reduced, and the image optimization time of the magnetic resonance image based on the second nuclide can be shortened, thereby improving the optimization efficiency of the magnetic resonance image to be optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 A schematic flow chart of an embodiment of the magnetic resonance image optimization method provided by the present invention;

[0045] Figure 2 For the present invention Figure 1 A schematic flow chart of an embodiment of S102;

[0046] Figure 3 For the present invention Figure 2 A schematic flow chart of an embodiment of S203;

[0047] Figure 4 For the present invention Figure 3 A schematic flow chart of an embodiment of S303;

[0048] Figure 5 A schematic diagram of a flow chart of an embodiment of preprocessing a magnetic resonance image sample provided by the present invention;

[0049] Figure 6 A schematic diagram of a flow chart of an embodiment of testing the second magnetic resonance image optimization model provided by the present invention;

[0050] Figure 7 A schematic diagram of the structure of an embodiment of a magnetic resonance image optimization device provided by the present invention;

[0051] Figure 8 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention. 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0053] In the description of the embodiments of the present invention, unless otherwise specified, "multiple" means two or more than two. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone.

[0054] The descriptions of "first", "second", etc. involved in the embodiments of the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the technical features defined as "first" or "second" may explicitly or implicitly include at least one of the features.

[0055] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0056] Before presenting the embodiments, the following terms are explained.

[0057] Concentration: refers to the proportion of a certain nuclide in the organism to all nuclides in the organism.

[0058] Relaxation: In the nuclear magnetic resonance phenomenon, relaxation refers to the phenomenon that when the atomic nucleus is resonant and in a high-energy state, it will quickly recover to its original low-energy state after the radio frequency pulse stops. The recovery process is called relaxation.

[0059] Transverse relaxation time: The relaxation process is divided into two steps, namely, the longitudinal magnetization intensity vector recovers to the initial equilibrium state and the transverse magnetization intensity decays to zero. The process of the transverse magnetization intensity gradually decaying from the maximum to zero is called the transverse relaxation process, and the time required for this process is the transverse relaxation time.

[0060] Gyromagnetic ratio: In the magnetic resonance phenomenon, the gyromagnetic ratio refers to the ratio of the angular frequency of an atom undergoing Larmor precession in a magnetic field to the magnetic induction intensity.

[0061] The present invention provides a magnetic resonance image optimization method, device, electronic device and storage medium, which are described below respectively.

[0062] Figure 1 A schematic flow chart of an embodiment of the magnetic resonance image optimization method provided by the present invention is shown in FIG. Figure 1 As shown, the magnetic resonance image optimization method includes:

[0063] S101, acquiring a first magnetic resonance image optimization model based on a first nuclear species;

[0064] S102, constructing a second magnetic resonance image optimization model according to the first magnetic resonance image optimization model based on a deep transfer learning algorithm;

[0065] S103, acquiring a magnetic resonance image to be optimized based on the second nuclear species, and optimizing the magnetic resonance image to be optimized by using a second magnetic resonance image optimization model;

[0066] The concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

[0067] Compared with the prior art, the magnetic resonance image optimization method provided by the embodiment of the present invention first obtains a first magnetic resonance image optimization model based on a first nuclide with a large concentration, a long transverse relaxation time and a large gyromagnetic ratio in a biological body, and then builds a second magnetic resonance image optimization model based on the first magnetic resonance image optimization model based on a deep transfer learning algorithm, and optimizes the magnetic resonance image to be optimized generated by the second nuclide with a low concentration, a short transverse relaxation time and a small gyromagnetic ratio in a biological body through the second magnetic resonance image optimization model, without modifying the existing magnetic resonance equipment, and therefore, without increasing the cost of optimizing the magnetic resonance image to be optimized. Furthermore, since the embodiment of the present invention constructs the second magnetic resonance image optimization model based on the deep transfer learning algorithm, the number of magnetic resonance images based on the second nuclide required to construct the second magnetic resonance image optimization model can be greatly reduced, and the image optimization time of the magnetic resonance image based on the second nuclide can be shortened, thereby improving the optimization efficiency of the magnetic resonance image to be optimized.

[0068] It should be noted that the first magnetic resonance image optimization model in step S101 can be constructed based on a convolutional neural network (CNN) model, wherein the convolutional neural network model can be any one of the convolutional neural network models such as the LetNet model, the AlexNet model, the VGGNet model, the ResNet model, the Inception model, and the Xception model.

[0069] The organism in the embodiment of the present invention may be a human body or an animal.

[0070] In some embodiments of the present invention, the first nuclide is 1 H, the second nuclide is 23 Nah.

[0071] because 23 Na plays an important role in maintaining cell metabolism and water balance in organisms. 23 Na is very sensitive to diseases and can be used to indicate the integrity of cells and metabolism. Therefore, by setting the second isotope to 23Na can further improve the research and application value of the magnetic resonance images to be optimized, and can improve the effectiveness of magnetic resonance images in disease diagnosis. 1 The MRI technology of H is relatively mature. Therefore, by setting the first isotope to 1 H can acquire a large number of first magnetic resonance image sets at a relatively low cost, thereby further reducing the cost of optimizing the magnetic resonance images.

[0072] It should be noted that the selection of the second nuclide can be adjusted according to actual needs and is not limited to 23 Well, I won’t go into details here.

[0073] In some embodiments of the present invention, Figure 2 As shown, step S102 includes:

[0074] S201, obtaining a model structure and model parameters of a first magnetic resonance image optimization model;

[0075] S202, based on the deep transfer learning algorithm, construct a migration model according to the model structure and model parameters;

[0076] S203, obtaining a second magnetic resonance image set based on a second nuclear species, training the migration model through the second magnetic resonance image set, and obtaining a second magnetic resonance image optimization model.

[0077] Among them, the model parameters in step S201 may be parameters of each layer such as the convolution layer and the pooling layer in the first magnetic resonance image optimization model, for example: the convolution kernel size and the weight of each layer, etc., which will not be described one by one here.

[0078] The model structure in step S201 may be the model structure of any one of the LetNet model, AlexNet model, VGGNet model, ResNet model, Inception model, and Xception model.

[0079] The migration model in step S202 refers to a model whose model structure and model parameters are the same as those of the first magnetic resonance image optimization model.

[0080] It should be understood that the second magnetic resonance image set in step S203 includes a plurality of second magnetic resonance images with different resolutions and signal-to-noise ratios, wherein the second magnetic resonance image based on the second nuclear species can be obtained by repeated sampling by a magnetic resonance device.

[0081] The embodiment of the present invention can greatly reduce the number of second magnetic resonance image sets required to obtain the second magnetic resonance image optimization model based on a deep transfer learning algorithm and by utilizing the model structure and model parameters of the first resonance image optimization model, thereby improving the optimization efficiency of the magnetic resonance image to be optimized.

[0082] In some embodiments of the present invention, Figure 3 As shown, step S203 includes:

[0083] S301, selecting a plurality of magnetic resonance image samples from a plurality of second magnetic resonance images;

[0084] S302, inputting the magnetic resonance image samples into the migration model for training to obtain training magnetic resonance images;

[0085] S303, determining whether the training of the migration model is complete based on the training magnetic resonance image and the magnetic resonance image samples;

[0086] S304: When the migration model has been trained completely, the trained migration model is used as the second magnetic resonance image optimization model.

[0087] The ratio of the number of magnetic resonance image samples to the number of the second magnetic resonance images in step S301 includes but is not limited to 0.8.

[0088] In some embodiments of the present invention, the training magnetic resonance image includes the current training resonance sub-image and the last training resonance sub-image. Figure 4 As shown, step S303 includes:

[0089] S401, determining a first peak signal-to-noise ratio between a last training resonance sub-image and a magnetic resonance image sample;

[0090] S402, determining a second peak signal-to-noise ratio between the current training resonance sub-image and the magnetic resonance image sample;

[0091] S403, determining whether the difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than a threshold difference;

[0092] S404: If the difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than the threshold difference, the migration model has been fully trained.

[0093] Among them, the calculation formula of the first peak signal-to-noise ratio PSNR is:

[0094]

[0095] In the formula, MAX Iis the maximum value of the color of the image point in the last training resonance sub-image and magnetic resonance image samples; MSE is the mean square error of the last training resonance sub-image and magnetic resonance image samples.

[0096] The calculation method of the second peak signal-to-noise ratio is the same as the calculation formula of the first peak signal-to-noise ratio, which will not be described in detail here.

[0097] It should be understood that when the difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than the threshold difference, it indicates that the difference between the results of the two trainings is not large, that is, the training of the migration model is complete.

[0098] In order to avoid unnecessary image information in the second magnetic resonance image, which affects the training speed when training the migration model, thereby reducing the construction speed of the second magnetic resonance optimization model and further reducing the optimization efficiency of the magnetic resonance image to be optimized, in some embodiments of the present invention, such as Figure 5 As shown, after step S301, the following steps are also included:

[0099] S501, performing grayscale transformation on the magnetic resonance image sample to obtain a grayscale image;

[0100] S502, performing histogram equalization processing on the grayscale image to obtain a balanced image;

[0101] S503: Segment and calibrate the equalized image using an edge detection algorithm to obtain a valid image.

[0102] The embodiment of the present invention can improve the image quality of the magnetic resonance image artifact by performing grayscale transformation on the magnetic resonance image samples; and can improve the contrast of the equalized image by performing histogram equalization processing on the grayscale image; that is, by performing grayscale transformation and histogram equalization processing on the magnetic resonance image samples in sequence, the magnetic resonance image samples that were originally blurred or even indistinguishable can be transformed into clear usable images rich in a large amount of useful information, effectively removing noise in the magnetic resonance image samples, and further improving the construction speed of the second magnetic resonance image optimization model.

[0103] Furthermore, the embodiment of the present invention performs segmentation and calibration on the equalized image through an edge detection algorithm, which can remove non-interest regions in the equalized image and retain the interest regions, thereby making it easier to train the interest regions in the magnetic resonance image samples, reducing the impact of non-interest regions on the training process, further improving the construction speed of the second magnetic resonance image optimization model, and improving the accuracy of the second magnetic resonance image optimization model in optimizing the magnetic resonance image to be optimized.

[0104] It should be noted that the grayscale transformation in step S501 may be at least one of image inversion, logarithmic transformation, gamma transformation and the like.

[0105] It should also be noted that the edge detection algorithm in step S503 may be a Mask RCNN algorithm.

[0106] In some embodiments of the present invention, in order to avoid the second magnetic resonance image optimization model having a low generalization ability, which results in the second magnetic resonance image optimization model being unable to effectively optimize the magnetic resonance image to be optimized, in some embodiments of the present invention, Figure 6 As shown, after step S304, the following further includes:

[0107] S601, selecting a plurality of magnetic resonance test images from a plurality of second magnetic resonance images;

[0108] S602: Verify the model performance of the second magnetic resonance image optimization model based on the magnetic resonance test image.

[0109] The ratio of the number of magnetic resonance image samples to the number of the plurality of second magnetic resonance test images in step S601 includes but is not limited to 0.2.

[0110] The embodiment of the present invention verifies the model performance such as the generalization ability of the second magnetic resonance image optimization model by selecting a magnetic resonance test image, thereby ensuring the optimization performance of the second magnetic resonance image optimization model for the magnetic resonance image to be optimized.

[0111] It should be noted that: in step S101, obtaining the first magnetic resonance image optimization model based on the first nuclide can first obtain a first magnetic resonance image set based on the first nuclide, and construct a first magnetic resonance image optimization model based on the first magnetic resonance image set, and the process of constructing the first magnetic resonance image optimization model based on the first magnetic resonance image set is the same as the process of training the migration model based on the second magnetic resonance image set to obtain the second magnetic resonance image optimization model, which will not be described in detail here.

[0112] In order to further improve the efficiency of constructing the second magnetic resonance image optimization model, in some embodiments of the present invention, when constructing the first magnetic resonance image optimization model in step S101, the first magnetic resonance image set can be subjected to grayscale transformation and histogram equalization processing, and the first magnetic resonance image set can be segmented and calibrated by an edge detection algorithm.

[0113] The embodiment of the present invention can improve the efficiency of constructing the first magnetic resonance image optimization model by performing grayscale transformation and histogram equalization processing on the first magnetic resonance image set and segmenting and calibrating the first magnetic resonance image set through an edge detection algorithm, thereby further improving the efficiency of constructing the second magnetic resonance image optimization model.

[0114] In order to better implement the magnetic resonance image optimization method in the embodiment of the present invention, based on the magnetic resonance image optimization method, correspondingly, Figure 7As shown, the embodiment of the present invention further provides a magnetic resonance image optimization device, and the magnetic resonance image optimization device 700 includes:

[0115] A first optimization model building unit 701 is used to obtain a first magnetic resonance image optimization model based on a first nuclide;

[0116] A second optimization model construction unit 702 is configured to construct a second magnetic resonance image optimization model according to the first magnetic resonance image optimization model based on a deep transfer learning algorithm;

[0117] The magnetic resonance image optimization unit 703 is used to obtain the magnetic resonance image to be optimized based on the second nuclear species, and optimize the magnetic resonance image to be optimized by using the second magnetic resonance image optimization model;

[0118] The concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

[0119] The magnetic resonance image optimization device 700 provided in the above embodiment can implement the technical solution described in the above magnetic resonance image optimization method embodiment. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above magnetic resonance image optimization method embodiment, which will not be repeated here.

[0120] like Figure 8 As shown, the present invention also provides an electronic device 800. The electronic device 800 includes a processor 801, a memory 802 and a display 803. Figure 8 Only some components of the electronic device 800 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0121] In some embodiments, the processor 801 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 802, such as the magnetic resonance image optimization method of the present invention.

[0122] In some embodiments, the processor 801 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, the processor 801 may be local or remote. In some embodiments, the processor 801 may be implemented in a cloud platform. In one embodiment, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0123] In some embodiments, the memory 802 may be an internal storage unit of the electronic device 800, such as a hard disk or memory of the electronic device 800. In other embodiments, the memory 802 may also be an external storage device of the electronic device 800, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 800.

[0124] Furthermore, the memory 802 may include both an internal storage unit of the electronic device 800 and an external storage device. The memory 802 is used to store application software installed in the electronic device 800 and various data.

[0125] In some embodiments, the display 803 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 803 is used to display information on the electronic device 800 and to display a visual user interface. The components 801-803 of the electronic device 800 communicate with each other via a system bus.

[0126] In one embodiment, when the processor 801 executes the magnetic resonance image optimization program in the memory 802, the following steps may be implemented:

[0127] acquiring a first magnetic resonance image optimization model based on a first nuclear species;

[0128] Based on the deep transfer learning algorithm, a second magnetic resonance image optimization model is constructed according to the first magnetic resonance image optimization model;

[0129] Acquiring a magnetic resonance image to be optimized based on a second nuclear species, and optimizing the magnetic resonance image to be optimized by using a second magnetic resonance image optimization model;

[0130] The concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

[0131] It should be understood that: when the processor 801 executes the magnetic resonance image optimization program in the memory 802, in addition to the above functions, other functions may also be implemented, and the details may refer to the description of the above corresponding method embodiments.

[0132] Furthermore, the embodiment of the present invention does not specifically limit the type of the electronic device 800 mentioned, and the electronic device 800 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (PDA), a wearable device, a laptop computer, etc. Exemplary embodiments of portable electronic devices include but are not limited to portable electronic devices equipped with IOS, Android, Microsoft or other operating systems. The above-mentioned portable electronic device may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 800 may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0133] Accordingly, an embodiment of the present application also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, the steps or functions in the magnetic resonance image optimization method provided by the above-mentioned method embodiments can be implemented.

[0134] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0135] The magnetic resonance image optimization method, device, electronic device and storage medium provided by the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A magnetic resonance image optimization method, characterized in that: include: acquiring a first magnetic resonance image optimization model based on a first nuclear species; Acquiring a model structure and model parameters of the first magnetic resonance image optimization model; Based on the deep transfer learning algorithm, construct a transfer model according to the model structure and the model parameters; Acquiring a second magnetic resonance image set based on a second nuclear species, training the migration model through the second magnetic resonance image set, and obtaining a second magnetic resonance image optimization model; Acquiring a magnetic resonance image to be optimized based on a second nuclear species, and optimizing the magnetic resonance image to be optimized by using the second magnetic resonance image optimization model; Wherein, the concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

2. The magnetic resonance image optimization method according to claim 1, characterized in that: The first nuclide is 1 H, the second nuclide is 23 Nah.

3. The magnetic resonance image optimization method according to claim 1, characterized in that: The second magnetic resonance image set includes a plurality of second magnetic resonance images; and the step of training the migration model through the second magnetic resonance image set to obtain the second magnetic resonance image optimization model includes: selecting a plurality of magnetic resonance image samples from the plurality of second magnetic resonance images; Inputting the magnetic resonance image samples into the migration model for training to obtain training magnetic resonance images; determining whether training of the migration model is complete based on the training magnetic resonance images and the magnetic resonance image samples; When the migration model has been fully trained, the fully trained migration model is the second magnetic resonance image optimization model.

4. The magnetic resonance image optimization method according to claim 3, characterized in that: The training magnetic resonance image includes a current training resonance sub-image and a previous training resonance sub-image; and determining whether the training of the migration model is complete based on the training magnetic resonance image and the magnetic resonance image sample includes: determining a first peak signal-to-noise ratio between the last training resonance sub-image and the magnetic resonance image sample; determining a second peak signal-to-noise ratio between the current training resonance subimage and the magnetic resonance image sample; Determining whether a difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than a threshold difference; If the difference between the first peak signal-to-noise ratio and the second peak signal-to-noise ratio is less than a threshold difference, the migration model has been fully trained.

5. The magnetic resonance image optimization method according to claim 3, characterized in that: After selecting a plurality of magnetic resonance image samples from the plurality of second magnetic resonance images, the method further includes: Performing grayscale transformation on the magnetic resonance image sample to obtain a grayscale image; Performing histogram equalization processing on the grayscale image to obtain a balanced image; The equalized image is segmented and calibrated by an edge detection algorithm to obtain a valid image.

6. The magnetic resonance image optimization method according to claim 3, characterized in that: After the migration model has been trained, it also includes: Selecting a plurality of magnetic resonance test images from the plurality of second magnetic resonance images; The model performance of the second magnetic resonance image optimization model is verified based on the magnetic resonance test image.

7. A magnetic resonance image optimization device, characterized in that: include: A first optimization model building unit, used for acquiring a first magnetic resonance image optimization model based on a first nuclear species; a second optimization model building unit, configured to obtain a model structure and model parameters of the first magnetic resonance image optimization model; Based on the deep transfer learning algorithm, construct a transfer model according to the model structure and the model parameters; Acquiring a second magnetic resonance image set based on a second nuclear species, training the migration model through the second magnetic resonance image set, and obtaining a second magnetic resonance image optimization model; a magnetic resonance image optimization unit, configured to obtain a magnetic resonance image to be optimized based on a second nuclear species, and optimize the magnetic resonance image to be optimized by using the second magnetic resonance image optimization model; Wherein, the concentration of the first nuclide in the organism is greater than the concentration of the second nuclide in the organism, and / or the transverse relaxation time of the first nuclide is greater than the transverse relaxation time of the second nuclide, and / or the gyromagnetic ratio of the first nuclide is greater than the gyromagnetic ratio of the second nuclide.

8. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the magnetic resonance image optimization method described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps of the magnetic resonance image optimization method described in any one of claims 1 to 6.

Images