Magnetic resonance imaging method, apparatus, device, and storage medium

Abstract

The application relates to a magnetic resonance imaging method, device, equipment and storage medium. The method comprises the following steps: acquiring a magnetic resonance signal of a detection object; the magnetic resonance signal comprises amplitude information and phase information; the phase information comprises interference phase information; a target amplitude graph and a target phase graph are generated according to the amplitude information and the phase information respectively; an image reconstruction model is used to perform image reconstruction processing according to the target amplitude graph and the target phase graph, so that a target real part graph is obtained; and the target real part graph corresponds to the magnetic resonance signal with suppressed interference phase information. According to the method, the polarity of a magnetization vector can be correctly judged, and a real part graph with correct contrast can be obtained.

Description

Technical Field

[0001] This application relates to the field of magnetic resonance technology, and in particular to a magnetic resonance imaging method, apparatus, device, and storage medium. Background Technology

[0002] Magnetic Resonance Imaging (MRI) is an imaging technique that uses signals generated by the resonance of atomic nuclei in a strong magnetic field to reconstruct images.

[0003] In related technologies, some information from the magnetic resonance signal is encoded into the phase, thereby enabling technologies such as temperature imaging, elastography, cardiac PSIR imaging, and 3D-real IR real part imaging for the diagnosis of Meniere's disease of the inner ear.

[0004] However, the phase information obtained by magnetic resonance imaging is often affected by a variety of factors, such as changes in magnetic susceptibility, eddy currents, chemical shift, and associated field effects. These factors can interfere with the actual phase of the magnetic resonance signal, leading to misjudgment of the polarity of the magnetization vector in the real part reconstruction, and thus failing to obtain a real part image with correct contrast. Summary of the Invention

[0005] Therefore, it is necessary to provide a magnetic resonance imaging method, apparatus, device, and storage medium that can avoid misjudging the polarity of the magnetization vector in real part reconstruction and thus obtain a real part image with correct contrast, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a magnetic resonance imaging method. The method includes:

[0007] Acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information;

[0008] Generate the target amplitude map and the target phase map based on the amplitude information and phase information, respectively.

[0009] The image reconstruction model is pre-trained to perform image reconstruction based on the target amplitude map and the target phase map to obtain the target real part map; wherein the target real part map corresponds to the magnetic resonance signal with suppressed interference phase information.

[0010] In one embodiment, the above-described image reconstruction process using a pre-trained image reconstruction model to obtain a target real part image based on the target magnitude map and the target phase map includes:

[0011] The target amplitude map and target phase map are input into the image reconstruction model to obtain intermediate phase information. Based on the intermediate phase information and the target amplitude map, image reconstruction is performed to output the target real part map. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase map.

[0012] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model. The above-mentioned image reconstruction model, which is pre-trained, performs image reconstruction processing based on the target magnitude map and the target phase map to obtain the target real part map, including:

[0013] The target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain the first interference phase map output by the first reconstruction sub-model after filtering out the interference phase information from the phase information based on the target phase map and performing image reconstruction processing based on the interference phase information and the target amplitude map.

[0014] The target amplitude map, target phase map, and first interference phase map are input into the second reconstruction sub-model to obtain intermediate phase information by suppressing interference phase information based on the first interference phase map and the target phase map. The target real part map is then output by performing image reconstruction processing based on the intermediate phase information and the target amplitude map.

[0015] In one embodiment, the method further includes:

[0016] Multiple training samples and the corresponding gold standard for each training sample are obtained; the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0017] The neural network model is trained based on multiple training samples and the gold standard corresponding to each training sample to obtain the image reconstruction model.

[0018] In one embodiment, obtaining multiple training samples and the gold standard corresponding to each training sample includes:

[0019] Acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive;

[0020] The interference phase information of the samples is determined based on the magnetic resonance signals of the first and second samples.

[0021] Based on the first sample magnetic resonance signal and the interference phase information of the sample, a sample amplitude map, a sample phase map, and a sample real part map are generated. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0022] In one embodiment, the acquisition of the first sample magnetic resonance signal and the second sample magnetic resonance signal includes:

[0023] The first scanning sequence and the second scanning sequence were used to scan, and the magnetic resonance signals of the first sample and the second sample were acquired.

[0024] The first scan sequence includes an inverted pulse, and the second scan sequence includes a saturated pulse.

[0025] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; the determination of the interference phase information of the sample based on the first sample magnetic resonance signal and the second sample magnetic resonance signal includes:

[0026] The first phase information and the second phase information are compared, and the interference phase information of the sample is determined based on the comparison result.

[0027] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, and the second sample magnetic resonance signal further includes second amplitude information. The generation of the sample amplitude map, sample phase map, and sample real part map based on the first sample magnetic resonance signal and the interference phase information of the sample includes:

[0028] The sample amplitude map and sample phase map are generated based on the first amplitude information and the first phase information, respectively, to obtain the training samples;

[0029] The intermediate phase information of the first sample magnetic resonance signal is obtained by suppressing the interfering phase information in the first phase information;

[0030] The gold standard corresponding to the training sample is generated based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0031] In one embodiment, the gold standard further includes a second interference phase map, and the acquisition of multiple training samples and the gold standard corresponding to each training sample includes:

[0032] A second interference phase map is generated based on the interference phase information of the sample;

[0033] Correspondingly, the above-mentioned training of the neural network model based on multiple training samples and the gold standard corresponding to each training sample yields an image reconstruction model, including:

[0034] The first reconstruction sub-model is trained based on multiple training samples and the second interference phase map corresponding to each training sample.

[0035] The second reconstruction sub-model is trained based on multiple training samples, the real part map of each training sample, and the multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples.

[0036] The image reconstruction model consists of the first reconstruction sub-model and the second reconstruction sub-model.

[0037] Secondly, this application also provides a magnetic resonance imaging device. The device includes:

[0038] The signal acquisition module is used to acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information.

[0039] The amplitude-phase diagram generation module is used to generate target amplitude diagrams and target phase diagrams based on amplitude information and phase information, respectively.

[0040] The real part image generation module is used to perform image reconstruction processing based on the target amplitude image and the target phase image using a pre-trained image reconstruction model to obtain the target real part image; wherein, the target real part image corresponds to the magnetic resonance signal with suppressed interference phase information.

[0041] In one embodiment, the real part image generation module is specifically used to input the target amplitude image and the target phase image into the image reconstruction model to obtain intermediate phase information, and to perform image reconstruction based on the intermediate phase information and the target amplitude image to output the target real part image. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase image.

[0042] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model. The aforementioned real part image generation module is specifically used to input the target amplitude image and the target phase image into the first reconstruction sub-model to obtain a first interference phase image output by the first reconstruction sub-model after filtering out interference phase information from the phase information based on the target phase image and performing image reconstruction processing based on the interference phase information and the target amplitude image; and to input the target amplitude image, the target phase image, and the first interference phase image into the second reconstruction sub-model to obtain a target real part image output by the second reconstruction sub-model after suppressing interference phase information based on the first interference phase image and the target phase image and performing image reconstruction processing based on the intermediate phase information and the target amplitude image.

[0043] In one embodiment, the device further includes:

[0044] The sample acquisition module is used to acquire multiple training samples and the gold standard corresponding to each training sample; among them, the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0045] The model training module is used to train a neural network model based on multiple training samples and the gold standard corresponding to each training sample, so as to obtain an image reconstruction model.

[0046] In one embodiment, the sample acquisition module includes:

[0047] The signal acquisition submodule is used to acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive.

[0048] The interference determination submodule is used to determine the interference phase information of the sample based on the magnetic resonance signal of the first sample and the magnetic resonance signal of the second sample.

[0049] The sample generation submodule is used to generate a sample amplitude map, a sample phase map, and a sample real part map based on the first sample magnetic resonance signal and the interference phase information of the sample. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0050] In one embodiment, the signal acquisition submodule is specifically used to perform scanning using a first scanning sequence and a second scanning sequence, and to acquire a first sample magnetic resonance signal and a second sample magnetic resonance signal; wherein the first scanning sequence includes an inversion pulse and the second scanning sequence includes a saturation pulse.

[0051] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; the interference determination submodule is specifically used to compare the first phase information and the second phase information, and determine the interference phase information of the sample based on the comparison result.

[0052] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, and the second sample magnetic resonance signal further includes second amplitude information. The sample generation submodule is specifically used to generate a sample amplitude map and a sample phase map based on the first amplitude information and the first phase information, respectively, to obtain training samples; to suppress the interfering phase information in the first phase information to obtain intermediate phase information of the first sample magnetic resonance signal; and to generate the gold standard corresponding to the training sample based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0053] In one embodiment, the gold standard further includes a second interference phase map, and the sample acquisition module is specifically used to generate the second interference phase map based on the interference phase information of the sample.

[0054] Correspondingly, the above-mentioned model training module is specifically used to train the first reconstruction sub-model based on multiple training samples and the second interference phase map corresponding to each training sample; to train the second reconstruction sub-model based on multiple training samples, the sample real part map corresponding to each training sample, and multiple third interference phase maps output by the first reconstruction sub-model according to multiple training samples; and to form an image reconstruction model by the first reconstruction sub-model and the second reconstruction sub-model.

[0055] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method described in the first aspect.

[0056] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the method described in the first aspect.

[0057] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the method described in the first aspect.

[0058] The aforementioned magnetic resonance imaging method, apparatus, device, and storage medium acquire the magnetic resonance signal of the target object; generate a target amplitude map and a target phase map based on amplitude information and phase information, respectively; and perform image reconstruction processing using a pre-trained image reconstruction model based on the target amplitude map and target phase map to obtain a target real part map. In this embodiment, the image reconstruction model suppresses interfering phase information in the phase information. Therefore, by generating the target real part map based on intermediate phase information, misjudgment of the polarity of the magnetization vector can be avoided in real part reconstruction, thereby obtaining a real part map with correct contrast. Attached Figure Description

[0059] Figure 1 This is a diagram illustrating the application environment of a magnetic resonance imaging method in one embodiment.

[0060] Figure 2 This is a schematic flowchart of a magnetic resonance imaging method in one embodiment;

[0061] Figure 3a This is one of the structural schematic diagrams of an image reconstruction model in one embodiment;

[0062] Figure 3b This is a second schematic diagram of the structure of an image reconstruction model in one embodiment;

[0063] Figure 4 This is a flowchart illustrating the image reconstruction process based on the target amplitude map and the target phase map in one embodiment.

[0064] Figure 5 This is one of the schematic diagrams illustrating the model training process in one embodiment;

[0065] Figure 6 This is a flowchart illustrating the steps for obtaining multiple training samples and the gold standard in one embodiment.

[0066] Figure 7 This is the second schematic diagram of the model training process in one embodiment;

[0067] Figure 8 This is a structural block diagram of a magnetic resonance imaging device in one embodiment;

[0068] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0070] The magnetic resonance imaging method provided in this application embodiment can be applied to, for example... Figure 1 The application environment shown may include computer device 101 and MR (Magnetic Resonance) device 102. Computer device 101 can communicate with MR device 102 via a network. Computer device 101 may be, but is not limited to, various personal computers, laptops, and tablets.

[0071] The application environment may also include a controller 103, which may include a processor and memory. The processor establishes a direct or indirect communicable connection with the computer device 101, the magnetic resonance device 102, and the memory via a bus. The controller 103 may be implemented using a standalone server or a server cluster consisting of multiple servers. The processor included in the controller 103 may be of any type, having one or more processing cores. It can perform single-threaded or multi-threaded operations, used for parsing instructions to perform operations such as acquiring data, performing logical operations, and distributing processing results. The memory included in the controller 103 may include a non-volatile computer-readable storage medium, such as at least one disk storage device, flash memory device, a distributed storage device remotely located relative to the processor, or other non-volatile solid-state storage devices.

[0072] In one embodiment, such as Figure 2 As shown, a magnetic resonance imaging method is provided, which is applied to... Figure 1 Taking a computer device as an example, the explanation includes the following steps:

[0073] Step 201: Obtain the magnetic resonance signal of the object being tested.

[0074] The above-mentioned magnetic resonance signal is a composite magnetic resonance signal, including amplitude information and phase information; the phase information includes interference phase information (background phase information) and actual phase information (phase information of the detected object). This magnetic resonance signal is shown in formula (1):

[0075]

[0076] Where ρ represents the amplitude information, This refers to phase information, which includes interference phase information. and actual phase information is the real part of the magnetic resonance signal. The imaginary part of the magnetic resonance signal is represented by , and the real part is used for image reconstruction. Interference phase information is also included. and actual phase information The sum is related to the spatial position of the magnetization vector and includes low-order and high-order, as well as linear and nonlinear background phases, for 2D imaging. 3D imaging x, y, z are the spatial coordinates of the reconstructed image.

[0077] Computer equipment can control magnetic resonance imaging (MRI) equipment to scan the object being tested, and then obtain the magnetic resonance signal of the object being tested from the MRI equipment.

[0078] Step 202: Generate the target amplitude map and the target phase map based on the amplitude information and phase information, respectively.

[0079] The computer device generates a target amplitude map based on the amplitude information and a target phase map based on the phase information. This embodiment does not limit the image generation algorithm.

[0080] Understandably, phase information includes both interference phase information and actual phase information; therefore, the generated target phase map also contains both interference phase information and actual phase information.

[0081] Step 203: Use a pre-trained image reconstruction model to perform image reconstruction processing based on the target amplitude map and the target phase map to obtain the target real part map.

[0082] Among them, the real part of the target corresponds to the magnetic resonance signal whose interference phase information is suppressed.

[0083] Computer equipment can pre-train an image reconstruction model. After generating the target amplitude map and the target phase map, the image reconstruction model is used to process the target phase map, suppressing the interfering phase information in the phase information to obtain intermediate phase information. Then, the image reconstruction model performs image reconstruction based on the intermediate phase information and the amplitude information contained in the target amplitude map to obtain the target real part map.

[0084] For example, the image reconstruction model determines the amplitude information ρ based on the target amplitude map and the phase information based on the target phase map. From phase information Remove some or all of the interfering phase information Intermediate phase information was then obtained. Then, based on the intermediate phase information The real part of the magnetic resonance signal can be determined by the amplitude information ρ. This generates the target real part diagram.

[0085] In the aforementioned magnetic resonance imaging method, the magnetic resonance signal of the object to be detected is acquired; a target amplitude map and a target phase map are generated based on the amplitude information and phase information, respectively; and a pre-trained image reconstruction model is used to perform image reconstruction processing based on the target amplitude map and the target phase map to obtain the target real part map. In this embodiment, the image reconstruction model suppresses interfering phase information in the phase information to obtain intermediate phase information. Therefore, generating the target real part map based on the intermediate phase information can avoid misjudging the polarity of the magnetization vector during real part reconstruction, thereby obtaining a real part map with correct contrast.

[0086] In one embodiment, the above-mentioned image reconstruction processing using a pre-trained image reconstruction model to obtain a target real part image based on a target amplitude map and a target phase map includes: inputting the target amplitude map and the target phase map into the image reconstruction model to obtain intermediate phase information, and performing image reconstruction based on the intermediate phase information and the target amplitude map to output the target real part image.

[0087] Among them, the intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase map.

[0088] The computer device inputs the target amplitude map and the target phase map into the image reconstruction model. The image reconstruction model can determine the amplitude information based on the target amplitude map and the phase information based on the target phase map. Then, the image reconstruction model suppresses the interfering phase information in the phase information to obtain intermediate phase information, and performs image reconstruction based on the intermediate phase information and amplitude information. Finally, it outputs the target real part map obtained from the image reconstruction.

[0089] The image reconstruction model described above can be a Convolutional Neural Network (CNN), which includes an input layer, hidden layers, and an output layer. The hidden layer includes at least one convolutional layer, at least one pooling layer, and a fully connected layer. For example... Figure 3a As shown, the hidden layer includes two convolutional layers and two pooling layers. This disclosure does not limit the structure of the convolutional neural network.

[0090] The input layer of a convolutional neural network (CNN) can process multidimensional data. After the target amplitude map and target phase map are input into the input layer of the image reconstruction model for processing, the convolutional layer extracts features from the processed target amplitude and phase maps. Then, the pooling layer selects and filters the features extracted by the convolutional layer, reducing the size of the feature maps. Since both the convolutional and pooling layers perform linear operations, and the interfering phase information often contains nonlinear information, an activation function is used to enable the CNN to learn nonlinear information. Refitting the nonlinear information through a fully connected layer reduces feature loss. Finally, the real part map of the target is output from the output layer.

[0091] In practical applications, a convolutional neural network of the appropriate dimension is used based on the dimensions of the target amplitude map and the target phase map. For example, if the target amplitude map and the target phase map are 2D images, a 2D convolutional neural network is used; if the target amplitude map and the target phase map are 3D images, a 3D convolutional neural network is used.

[0092] In the above embodiments, the target amplitude map and the target phase map are input into the image reconstruction model to obtain intermediate phase information. Based on the intermediate phase information and the target amplitude map, image reconstruction is performed to output the target real part map. Through the image reconstruction model in this embodiment, interfering phase information in the phase information can be suppressed relatively easily, resulting in more accurate intermediate phase information. Therefore, generating the target real part map based on the intermediate phase information avoids the problem of misjudging the polarity of the magnetization vector during real part reconstruction, thus obtaining a real part map with correct contrast when the polarity is correctly determined.

[0093] In one embodiment, such as Figure 3b As shown, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model. Figure 4 As shown, the process of using a pre-trained image reconstruction model to perform image reconstruction based on the target magnitude map and the target phase map to obtain the target real part map can include the following steps:

[0094] Step 301: Input the target amplitude map and the target phase map into the first reconstruction sub-model to obtain the first interference phase map output by the first reconstruction sub-model after filtering the interference phase information from the phase information according to the target phase map and performing image reconstruction processing according to the interference phase information and the target amplitude map.

[0095] The first reconstruction sub-model is used to extract interference phase information from the phase information.

[0096] The computer device inputs the target amplitude map and the target phase map into the first reconstruction sub-model. The first reconstruction sub-model determines the amplitude information based on the target amplitude map and the phase information based on the target phase map. Then, the first reconstruction sub-model filters out the interference phase information from the phase information and performs image reconstruction processing based on the interference phase information and amplitude information to output the first interference phase map.

[0097] Step 302: Input the target amplitude map, target phase map and first interference phase map into the second reconstruction sub-model to obtain the intermediate phase information by suppressing the interference phase information according to the first interference phase map and the target phase map, and then perform image reconstruction processing based on the intermediate phase information and the target amplitude map to output the target real part map.

[0098] The second reconstruction sub-model is used to determine intermediate phase information based on the interference phase information.

[0099] The computer device inputs the target amplitude map, the target phase map, and the first interference phase map into the second reconstruction sub-model. The second reconstruction sub-model determines the amplitude information based on the target amplitude map, the phase information based on the target phase map, and the interference phase information based on the first interference phase map. Then, the second reconstruction sub-model partially or completely removes the interference phase information from the phase information to obtain intermediate phase information. Next, the second reconstruction sub-model performs image reconstruction based on the intermediate phase information and the amplitude information, and outputs the target real part map.

[0100] In the above embodiments, the target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain a first interference phase map output by the first reconstruction sub-model after filtering out interference phase information from the phase information based on the target phase map and performing image reconstruction processing based on the interference phase information and the target amplitude map. The target amplitude map, the target phase map, and the first interference phase map are input into the second reconstruction sub-model to obtain an intermediate phase information obtained by the second reconstruction sub-model after suppressing interference phase information based on the first interference phase map and the target phase map and performing image reconstruction processing based on the intermediate phase information and the target amplitude map. In this embodiment, the interference phase information in the phase information is first extracted by the first reconstruction sub-model, and then the intermediate phase information is determined by the second reconstruction sub-module based on the extracted interference phase information. This determination method can improve the accuracy of the intermediate phase information, thereby correctly determining the polarity of the magnetization vector and obtaining a real part map with correct contrast.

[0101] In one embodiment, such as Figure 5 As shown, embodiments of this disclosure may further include a model training process, such as the following steps:

[0102] Step 401: Obtain multiple training samples and the gold standard corresponding to each training sample.

[0103] The training samples include sample amplitude maps and sample phase maps, while the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0104] Computer equipment can acquire multiple sample magnetic resonance signals from a magnetic resonance device, and generate corresponding sample amplitude maps and sample phase maps based on each sample magnetic resonance signal to obtain training samples.

[0105] The computer device partially or completely removes interfering phase information from the magnetic resonance signals of each sample, retains intermediate phase information, and generates the corresponding sample real part image based on the intermediate phase information to obtain the gold standard corresponding to the training sample. This disclosure does not limit the method of removing interfering phase information.

[0106] Step 402: Train the neural network model based on multiple training samples and the gold standard corresponding to each training sample to obtain the image reconstruction model.

[0107] A computer device inputs a training sample into a convolutional neural network (CNN), which outputs a corresponding training real part image based on the training sample. Then, the computer device calculates the loss value between the training real part image and the sample real part image using a preset loss function. If the loss value does not meet a preset convergence condition, the adjustable parameters in the CNN are adjusted, and another sample is input into the CNN to continue training. Training ends when the loss value between the training real part image and the sample real part image output by the CNN meets the preset convergence condition, and the CNN at the end of training is identified as the image reconstruction model. This embodiment does not limit the loss function and preset convergence condition; they can be set according to actual conditions.

[0108] In the above embodiments, multiple training samples and corresponding gold standards are obtained; a neural network model is trained based on the multiple training samples and corresponding gold standards to obtain an image reconstruction model. This embodiment trains an image reconstruction model so that it can be used in practical applications to remove interfering phase information. This not only improves the speed of determining intermediate phase information but also improves the accuracy of the intermediate phase information, thereby correctly determining the polarity of the magnetization vector and obtaining a real part image with correct contrast.

[0109] In one embodiment, such as Figure 6 As shown, the process of obtaining multiple training samples and the corresponding gold standard for each training sample can include the following steps:

[0110] Step 501: Obtain the first sample magnetic resonance signal and the second sample magnetic resonance signal.

[0111] The polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive.

[0112] The computer equipment controls the magnetic resonance equipment to perform scanning using the first scanning sequence and the second scanning sequence, and acquires the magnetic resonance signals of the first sample and the second sample.

[0113] The first scan sequence mentioned above includes inverted pulses, and the second scan sequence mentioned above includes saturated pulses. Examples include 3D-real IR real part imaging sequences used for the diagnosis of Meniere's disease of the inner ear, and cardiac PSIR imaging sequences.

[0114] For example, a computer device controls an MRI scanner to perform a scan using a first scan sequence including inverted pulses and acquire a first sample MRI signal. Then, the computer device controls the MRI scanner to perform a scan using a second scan sequence including saturation pulses and acquire a second sample MRI signal. Alternatively, the computer device controls the MRI scanner to perform a scan using a second scan sequence with an inversion time greater than a preset duration and acquire a second sample MRI signal.

[0115] Understandably, the first scan sequence includes a reversal pulse, and the polarity of the magnetization vector corresponding to the first sample magnetic resonance signal changes from negative to positive. The second scan sequence includes a saturation pulse, and the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive. Alternatively, the reversal time of the second scan sequence is greater than a preset duration, allowing sufficient time for the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal to turn positive. Because the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive, the influence of a negative magnetization vector polarity can be avoided, thus obtaining accurate phase information.

[0116] Step 502: Determine the interference phase information of the sample based on the magnetic resonance signal of the first sample and the magnetic resonance signal of the second sample.

[0117] The first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample.

[0118] After the computer equipment acquires the first sample magnetic resonance signal and the second sample magnetic resonance signal, it compares the first phase information and the second phase information, and determines the interference phase information of the sample based on the comparison result.

[0119] Understandably, the first sample magnetic resonance signal and the second sample magnetic resonance signal were acquired under the same conditions. Therefore, the interference phase information of the sample included in the first phase information is the same as the interference phase information of the sample included in the second phase information. After comparing the first phase information and the second phase information, the identical part between the first phase information and the second phase information can be identified as the interference information of the sample.

[0120] Step 503: Generate a sample amplitude map, a sample phase map, and a sample real part map based on the first sample magnetic resonance signal and the interference phase information of the sample. Use the sample amplitude map and the sample phase map as training samples, and use the sample real part map as the gold standard corresponding to the training samples.

[0121] The first sample magnetic resonance signal also includes first amplitude information, and the second sample magnetic resonance signal also includes second amplitude information.

[0122] The computer device generates sample amplitude maps and sample phase maps based on the first amplitude information and the first phase information, respectively, to obtain training samples; it suppresses the interfering phase information in the first phase information to obtain the intermediate phase information of the first sample magnetic resonance signal; and it generates the gold standard corresponding to the training samples based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0123] In the above embodiments, a first sample magnetic resonance signal and a second sample magnetic resonance signal are acquired; the interference phase information of the sample is determined based on the first sample magnetic resonance signal and the second sample magnetic resonance signal; a sample amplitude map, a sample phase map, and a sample real part map are generated based on the first sample magnetic resonance signal and the sample interference phase information, and the sample amplitude map and sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training sample. This embodiment extracts the interference phase information of the sample from the first sample magnetic resonance signal and the second sample magnetic resonance signal acquired under the same conditions, thereby generating a sample real part map that suppresses interference phase information based on the sample's interference phase information, providing a training basis for the image reconstruction model. Furthermore, since the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive, the influence of the magnetization vector polarity can be eliminated, resulting in more accurate second phase information. Thus, the interference phase information of the sample determined based on the second phase information is more accurate, thereby generating a gold standard with higher accuracy and improving the accuracy of the image reconstruction model.

[0124] In one embodiment, when the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model, the gold standard includes not only the sample real part map but also a second interference phase map. The process of obtaining multiple training samples and the gold standard corresponding to each training sample may include: generating a second interference phase map based on the interference phase information of the samples.

[0125] After determining the interference phase information of the sample based on the first phase information and the second phase information, the computer device generates a second interference phase map based on the interference phase information of the sample, and uses the second interference phase map as the gold standard used by the first reconstruction sub-model.

[0126] like Figure 7 As shown, the process of training a neural network model based on multiple training samples and the gold standard corresponding to each training sample to obtain an image reconstruction model can include the following steps:

[0127] Step 4021: Train the first reconstruction sub-model based on multiple training samples and the second interference phase map corresponding to each training sample.

[0128] The computer device inputs a training sample into a first convolutional neural network (CNN), which outputs a third interference phase map based on the training sample. Then, the computer device calculates a first loss value between the third interference phase map and the second interference phase map using a preset first loss function. If the first loss value does not meet a first preset convergence condition, the adjustable parameters in the first CNN are adjusted, and another training sample is input into the first CNN to continue training. Training of the first CNN ends when the computer device determines that the first loss value meets the first preset convergence condition.

[0129] Step 4022: Based on multiple training samples, the real part map of each training sample, and the multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples, the second reconstruction sub-model is trained.

[0130] While training the first reconstruction sub-model, the training samples input to the first convolutional neural network (CNN) and the third interference phase map output by the first CNN are input to the second CNN, which outputs a training real part map. Then, the computer calculates a second loss value between the training real part map and the sample real part map using a preset second loss function. If the second loss value does not meet the second preset convergence condition, the adjustable parameters in the second CNN are adjusted, and another training sample input to the first CNN is input into the second CNN to continue training. Training of the second CNN ends when the computer determines that the second loss value meets the second preset convergence condition.

[0131] Understandably, the first loss function and the second loss function can be the same or different loss functions; the first preset convergence condition and the second preset convergence condition can be the same or different preset convergence conditions. This disclosure does not limit the loss function or the preset convergence condition.

[0132] Step 4023: The image reconstruction model is composed of the first reconstruction sub-model and the second reconstruction sub-model.

[0133] The computer device terminates model training when it determines that the first loss value meets the first preset convergence condition and the second loss value meets the second preset convergence condition. The first convolutional neural network at the end of training is determined as the first reconstruction sub-model, and the second convolutional neural network at the end of training is determined as the second reconstruction sub-model. The first reconstruction sub-model and the second reconstruction sub-model constitute the image reconstruction model.

[0134] In the above embodiments, a first reconstruction sub-model is trained based on multiple training samples and the second interference phase map corresponding to each training sample; a second reconstruction sub-model is trained based on multiple training samples, the real part map of each training sample, and multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples; the first reconstruction sub-model and the second reconstruction sub-model constitute an image reconstruction model. This embodiment extracts interference phase information by training the first reconstruction sub-model, thus obtaining more accurate interference phase information, which in turn allows for more accurate intermediate phase information, enabling accurate determination of the polarity of the magnetization vector and obtaining a real part map with correct contrast.

[0135] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0136] Based on the same inventive concept, this application also provides a magnetic resonance imaging apparatus for implementing the magnetic resonance imaging method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more magnetic resonance imaging apparatus embodiments provided below can be found in the limitations of the magnetic resonance imaging method described above, and will not be repeated here.

[0137] In one embodiment, such as Figure 8 As shown, a magnetic resonance imaging device is provided, comprising:

[0138] The signal acquisition module 601 is used to acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information.

[0139] The amplitude-phase diagram generation module 602 is used to generate a target amplitude diagram and a target phase diagram based on the amplitude information and phase information, respectively.

[0140] The real part image generation module 603 is used to perform image reconstruction processing based on the target amplitude image and the target phase image using a pre-trained image reconstruction model to obtain the target real part image; wherein, the target real part image corresponds to the magnetic resonance signal whose interference phase information is suppressed.

[0141] In one embodiment, the real part image generation module 603 is specifically used to input the target amplitude image and the target phase image into the image reconstruction model to obtain intermediate phase information, and to perform image reconstruction based on the intermediate phase information and the target amplitude image to output the target real part image. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase image.

[0142] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model. The real part image generation module 603 is specifically used to input the target amplitude image and the target phase image into the first reconstruction sub-model to obtain a first interference phase image output by the first reconstruction sub-model after filtering out interference phase information from the phase information based on the target phase image and performing image reconstruction processing based on the interference phase information and the target amplitude image; and to input the target amplitude image, the target phase image and the first interference phase image into the second reconstruction sub-model to obtain a target real part image output by the second reconstruction sub-model after suppressing interference phase information based on the first interference phase image and the target phase image and performing image reconstruction processing based on the intermediate phase information and the target amplitude image.

[0143] In one embodiment, the device further includes:

[0144] The sample acquisition module 604 is used to acquire multiple training samples and the gold standard corresponding to each training sample; wherein, the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0145] The model training module 605 is used to train a neural network model based on multiple training samples and the gold standard corresponding to each training sample to obtain an image reconstruction model.

[0146] In one embodiment, the sample acquisition module 604 includes:

[0147] The signal acquisition submodule is used to acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive.

[0148] The interference determination submodule is used to determine the interference phase information of the sample based on the magnetic resonance signal of the first sample and the magnetic resonance signal of the second sample.

[0149] The sample generation submodule is used to generate a sample amplitude map, a sample phase map, and a sample real part map based on the first sample magnetic resonance signal and the interference phase information of the sample. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0150] In one embodiment, the signal acquisition submodule is specifically used to perform scanning using a first scanning sequence and a second scanning sequence, and to acquire a first sample magnetic resonance signal and a second sample magnetic resonance signal; wherein the first scanning sequence includes an inversion pulse and the second scanning sequence includes a saturation pulse.

[0151] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; the interference determination submodule is specifically used to compare the first phase information and the second phase information, and determine the interference phase information of the sample based on the comparison result.

[0152] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, and the second sample magnetic resonance signal further includes second amplitude information. The sample generation submodule is specifically used to generate a sample amplitude map and a sample phase map based on the first amplitude information and the first phase information, respectively, to obtain training samples; to suppress the interfering phase information in the first phase information to obtain intermediate phase information of the first sample magnetic resonance signal; and to generate the gold standard corresponding to the training sample based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0153] In one embodiment, the gold standard further includes a second interference phase map, and the sample acquisition module is specifically used to generate the second interference phase map based on the interference phase information of the sample.

[0154] Correspondingly, the above-mentioned model training module is specifically used to train the first reconstruction sub-model based on multiple training samples and the second interference phase map corresponding to each training sample; to train the second reconstruction sub-model based on multiple training samples, the sample real part map corresponding to each training sample, and multiple third interference phase maps output by the first reconstruction sub-model according to multiple training samples; and to form an image reconstruction model by the first reconstruction sub-model and the second reconstruction sub-model.

[0155] Each module in the aforementioned magnetic resonance imaging device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0156] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 9As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a magnetic resonance imaging method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0157] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0158] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0159] Acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information;

[0160] Generate the target amplitude map and the target phase map based on the amplitude information and phase information, respectively.

[0161] The image reconstruction model, which is pre-trained, is used to reconstruct the target image based on the target amplitude map and the target phase map to obtain the target real part map; the target real part map corresponds to the magnetic resonance signal whose interference phase information is suppressed.

[0162] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0163] The target amplitude map and target phase map are input into the image reconstruction model to obtain intermediate phase information. Based on the intermediate phase information and the target amplitude map, image reconstruction is performed to output the target real part map. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase map.

[0164] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model, and the processor, when executing the computer program, further implements the following steps:

[0165] The target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain the first interference phase map output by the first reconstruction sub-model after filtering out the interference phase information from the phase information based on the target phase map and performing image reconstruction processing based on the interference phase information and the target amplitude map.

[0166] The target amplitude map, target phase map, and first interference phase map are input into the second reconstruction sub-model to obtain intermediate phase information by suppressing interference phase information based on the first interference phase map and the target phase map. The target real part map is then output by performing image reconstruction processing based on the intermediate phase information and the target amplitude map.

[0167] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0168] Multiple training samples and the corresponding gold standard for each training sample are obtained; the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0169] The neural network model is trained based on multiple training samples and the gold standard corresponding to each training sample to obtain the image reconstruction model.

[0170] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0171] Acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive;

[0172] The interference phase information of the samples is determined based on the magnetic resonance signals of the first and second samples.

[0173] Based on the first sample magnetic resonance signal and the interference phase information of the sample, a sample amplitude map, a sample phase map, and a sample real part map are generated. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0174] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0175] The first scanning sequence and the second scanning sequence were used to scan, and the magnetic resonance signals of the first sample and the second sample were acquired.

[0176] The first scan sequence includes an inverted pulse, and the second scan sequence includes a saturated pulse.

[0177] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; the processor, when executing the computer program, further implements the following steps:

[0178] The first phase information and the second phase information are compared, and the interference phase information of the sample is determined based on the comparison result.

[0179] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, the second sample magnetic resonance signal further includes second amplitude information, and the processor, when executing the computer program, further implements the following steps:

[0180] The sample amplitude map and sample phase map are generated based on the first amplitude information and the first phase information, respectively, to obtain the training samples;

[0181] The intermediate phase information of the first sample magnetic resonance signal is obtained by suppressing the interfering phase information in the first phase information;

[0182] The gold standard corresponding to the training sample is generated based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0183] In one embodiment, the gold standard further includes a second interference phase map, and the processor, when executing the computer program, also performs the following steps:

[0184] A second interference phase map is generated based on the interference phase information of the sample;

[0185] The first reconstruction sub-model is trained based on multiple training samples and the second interference phase map corresponding to each training sample.

[0186] The second reconstruction sub-model is trained based on multiple training samples, the real part map of each training sample, and the multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples.

[0187] The image reconstruction model consists of the first reconstruction sub-model and the second reconstruction sub-model.

[0188] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0189] Acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information;

[0190] Generate the target amplitude map and the target phase map based on the amplitude information and phase information, respectively.

[0191] The image reconstruction model, which is pre-trained, is used to reconstruct the target image based on the target amplitude map and the target phase map to obtain the target real part map; the target real part map corresponds to the magnetic resonance signal whose interference phase information is suppressed.

[0192] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0193] The target amplitude map and target phase map are input into the image reconstruction model to obtain intermediate phase information. Based on the intermediate phase information and the target amplitude map, the target real part map is output by image reconstruction. The intermediate phase information is the phase information corresponding to the target phase map and is obtained by suppressing at least part of the interference phase information.

[0194] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model, and the computer program, when executed by a processor, further implements the following steps:

[0195] The target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain the first interference phase map output by the first reconstruction sub-model after filtering out the interference phase information from the phase information based on the target phase map and performing image reconstruction processing based on the interference phase information and the target amplitude map.

[0196] The target amplitude map, target phase map, and first interference phase map are input into the second reconstruction sub-model to obtain intermediate phase information by suppressing interference phase information based on the first interference phase map and the target phase map. The target real part map is then output by performing image reconstruction processing based on the intermediate phase information and the target amplitude map.

[0197] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0198] Multiple training samples and the corresponding gold standard for each training sample are obtained; the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0199] The neural network model is trained based on multiple training samples and the gold standard corresponding to each training sample to obtain the image reconstruction model.

[0200] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0201] Acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive;

[0202] The interference phase information of the samples is determined based on the magnetic resonance signals of the first and second samples.

[0203] Based on the first sample magnetic resonance signal and the interference phase information of the sample, a sample amplitude map, a sample phase map, and a sample real part map are generated. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0204] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0205] The first scanning sequence and the second scanning sequence were used to scan, and the magnetic resonance signals of the first sample and the second sample were acquired.

[0206] The first scan sequence includes an inverted pulse, and the second scan sequence includes a saturated pulse.

[0207] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; when the computer program is executed by the processor, it further implements the following steps:

[0208] The first phase information and the second phase information are compared, and the interference phase information of the sample is determined based on the comparison result.

[0209] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, the second sample magnetic resonance signal further includes second amplitude information, and the computer program, when executed by the processor, further implements the following steps:

[0210] The sample amplitude map and sample phase map are generated based on the first amplitude information and the first phase information, respectively, to obtain the training samples;

[0211] The intermediate phase information of the first sample magnetic resonance signal is obtained by suppressing the interfering phase information in the first phase information;

[0212] The gold standard corresponding to the training sample is generated based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0213] In one embodiment, the gold standard further includes a second interference phase map, and the computer program, when executed by a processor, also performs the following steps:

[0214] A second interference phase map is generated based on the interference phase information of the sample;

[0215] The first reconstruction sub-model is trained based on multiple training samples and the second interference phase map corresponding to each training sample.

[0216] The second reconstruction sub-model is trained based on multiple training samples, the real part map of each training sample, and the multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples.

[0217] The image reconstruction model consists of the first reconstruction sub-model and the second reconstruction sub-model.

[0218] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0219] Acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information;

[0220] Generate the target amplitude map and the target phase map based on the amplitude information and phase information, respectively.

[0221] The image reconstruction model, which is pre-trained, is used to reconstruct the target image based on the target amplitude map and the target phase map to obtain the target real part map; the target real part map corresponds to the magnetic resonance signal whose interference phase information is suppressed.

[0222] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0223] The target amplitude map and target phase map are input into the image reconstruction model to obtain intermediate phase information. Based on the intermediate phase information and the target amplitude map, the target real part map is output by image reconstruction. The intermediate phase information is the phase information corresponding to the target phase map and is obtained by suppressing at least part of the interference phase information.

[0224] In one embodiment, the image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model, and the computer program, when executed by a processor, further implements the following steps:

[0225] The target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain the first interference phase map output by the first reconstruction sub-model after filtering out the interference phase information from the phase information based on the target phase map and performing image reconstruction processing based on the interference phase information and the target amplitude map.

[0226] The target amplitude map, target phase map, and first interference phase map are input into the second reconstruction sub-model to obtain intermediate phase information by suppressing interference phase information based on the first interference phase map and the target phase map. The target real part map is then output by performing image reconstruction processing based on the intermediate phase information and the target amplitude map.

[0227] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0228] Multiple training samples and the corresponding gold standard for each training sample are obtained; the training samples include sample amplitude maps and sample phase maps, and the gold standard is the sample real part map that suppresses interference phase information and retains intermediate phase information.

[0229] The neural network model is trained based on multiple training samples and the gold standard corresponding to each training sample to obtain the image reconstruction model.

[0230] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0231] Acquire the first sample magnetic resonance signal and the second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive;

[0232] The interference phase information of the samples is determined based on the magnetic resonance signals of the first and second samples.

[0233] Based on the first sample magnetic resonance signal and the interference phase information of the sample, a sample amplitude map, a sample phase map, and a sample real part map are generated. The sample amplitude map and the sample phase map are used as training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

[0234] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0235] The first scanning sequence and the second scanning sequence were used to scan, and the magnetic resonance signals of the first sample and the second sample were acquired.

[0236] The first scan sequence includes an inverted pulse, and the second scan sequence includes a saturated pulse.

[0237] In one embodiment, the first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample; when the computer program is executed by the processor, it further implements the following steps:

[0238] The first phase information and the second phase information are compared, and the interference phase information of the sample is determined based on the comparison result.

[0239] In one embodiment, the first sample magnetic resonance signal further includes first amplitude information, the second sample magnetic resonance signal further includes second amplitude information, and the computer program, when executed by the processor, further implements the following steps:

[0240] The sample amplitude map and sample phase map are generated based on the first amplitude information and the first phase information, respectively, to obtain the training samples;

[0241] The intermediate phase information of the first sample magnetic resonance signal is obtained by suppressing the interfering phase information in the first phase information;

[0242] The gold standard corresponding to the training sample is generated based on the first amplitude information and the intermediate phase information of the first sample magnetic resonance signal.

[0243] In one embodiment, the gold standard further includes a second interference phase map, and the computer program, when executed by a processor, also performs the following steps:

[0244] A second interference phase map is generated based on the interference phase information of the sample;

[0245] The first reconstruction sub-model is trained based on multiple training samples and the second interference phase map corresponding to each training sample.

[0246] The second reconstruction sub-model is trained based on multiple training samples, the real part map of each training sample, and the multiple third interference phase maps output by the first reconstruction sub-model according to the multiple training samples.

[0247] The image reconstruction model consists of the first reconstruction sub-model and the second reconstruction sub-model.

[0248] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0249] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0250] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A magnetic resonance imaging method, characterized in that, The method includes: The magnetic resonance signal of the object to be detected is acquired; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information; the interference phase information includes linear and nonlinear background phase. Generate a target amplitude map and a target phase map based on the amplitude information and the phase information, respectively; An image reconstruction model, pre-trained, is used to reconstruct the target image based on the target amplitude map and the target phase map to obtain the target real part map; wherein, the target real part map corresponds to the magnetic resonance signal with suppressed interference phase information; The step of using a pre-trained image reconstruction model to perform image reconstruction processing based on the target magnitude map and the target phase map to obtain the target real part map includes: The target amplitude map and the target phase map are input into the image reconstruction model to obtain intermediate phase information. The target real part map is output by image reconstruction based on the intermediate phase information and the target amplitude map. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase map.

2. The method according to claim 1, characterized in that, The image reconstruction model includes a first reconstruction sub-model and a second reconstruction sub-model. The pre-trained image reconstruction model performs image reconstruction processing based on the target magnitude map and the target phase map to obtain the target real part map, including: The target amplitude map and the target phase map are input into the first reconstruction sub-model to obtain a first interference phase map output by the first reconstruction sub-model, which filters out the interference phase information from the phase information based on the target phase map, and performs image reconstruction processing based on the interference phase information and the target amplitude map. The target amplitude map, the target phase map, and the first interference phase map are input into the second reconstruction sub-model to obtain the target real part map by the second reconstruction sub-model suppressing the interference phase information according to the first interference phase map and the target phase map, obtaining intermediate phase information, and performing image reconstruction processing according to the intermediate phase information and the target amplitude map.

3. The method according to claim 1, characterized in that, The method further includes: Multiple training samples and a gold standard corresponding to each training sample are obtained; wherein, the training samples include sample amplitude maps and sample phase maps, and the gold standard is a sample real part map that suppresses interference phase information and retains intermediate phase information; The image reconstruction model is obtained by training a neural network model based on the multiple training samples and the gold standard corresponding to each training sample.

4. The method according to claim 3, characterized in that, The process of obtaining multiple training samples and the gold standard corresponding to each training sample includes: Acquire a first sample magnetic resonance signal and a second sample magnetic resonance signal; the polarity of the magnetization vector corresponding to the second sample magnetic resonance signal is positive; The interference phase information of the sample is determined based on the magnetic resonance signal of the first sample and the magnetic resonance signal of the second sample; The sample amplitude map, the sample phase map, and the sample real part map are generated based on the first sample magnetic resonance signal and the interference phase information of the sample. The sample amplitude map and the sample phase map are used as the training samples, and the sample real part map is used as the gold standard corresponding to the training samples.

5. The method according to claim 4, characterized in that, The acquisition of the first sample magnetic resonance signal and the second sample magnetic resonance signal includes: The first scanning sequence and the second scanning sequence were used to scan, and the magnetic resonance signals of the first sample and the second sample were acquired. The first scan sequence includes an inverted pulse, and the second scan sequence includes a saturated pulse.

6. The method according to claim 4, characterized in that, The first sample magnetic resonance signal includes first phase information, which includes interference phase information of the sample; the second sample magnetic resonance signal includes second phase information, which includes interference phase information of the sample. The step of determining the interference phase information of the sample based on the first sample magnetic resonance signal and the second sample magnetic resonance signal includes: The first phase information and the second phase information are compared, and the interference phase information of the sample is determined based on the comparison result.

7. A magnetic resonance imaging device, characterized in that, The device includes: The signal acquisition module is used to acquire the magnetic resonance signal of the object being detected; the magnetic resonance signal includes amplitude information and phase information; the phase information includes interference phase information; the interference phase information includes linear and nonlinear background phase; An amplitude-phase diagram generation module is used to generate a target amplitude diagram and a target phase diagram based on the amplitude information and the phase information, respectively. The real part image generation module is used to perform image reconstruction processing based on the target amplitude image and the target phase image using a pre-trained image reconstruction model to obtain the target real part image; wherein, the target real part image corresponds to the magnetic resonance signal with suppressed interference phase information; Specifically, the real part image generation module is used to input the target amplitude image and the target phase image into the image reconstruction model to obtain intermediate phase information, and to perform image reconstruction based on the intermediate phase information and the target amplitude image to output the target real part image. The intermediate phase information is obtained by suppressing at least part of the interference phase information corresponding to the target phase image.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images