Magnetic resonance imaging (MRI) scanning method, device and electronic equipment

Abstract

The application provides a magnetic resonance imaging (MRI) scanning method, device and electronic equipment. The method comprises the following steps: generating a first slice image of a current scanning slice based on original K-space data of the current scanning slice; determining whether the current scanning slice is a first type of scanning slice which is imaged by using an intra-frame imaging mode; in response to the current scanning slice not being the first type of scanning slice, determining a neighboring scanning slice corresponding to the current scanning slice and a second slice image corresponding to the current scanning slice and having a resolution greater than that of the first slice image; and generating a second slice image of the current scanning slice by using MR inter-frame prediction mode based on the first slice image and the second slice image of the neighboring scanning slice. In the application, the structural correlation between adjacent slice images is utilized to generate a slice image of each scanning slice, so that the purpose of high-speed and high-quality magnetic resonance MRI scanning and imaging is achieved.

Description

Technical Field

[0001] This application relates to the field of magnetic resonance imaging technology, and in particular to a scanning method, apparatus, electronic device, and computer-readable storage medium for magnetic resonance imaging (MRI). Background Technology

[0002] Magnetic resonance imaging (MRI), as an advanced imaging technique, has begun to be combined with surgical robots to enable complex surgical procedures, such as deep brain tumor ablation. Therefore, high-quality MRI imaging and high-speed imaging are required.

[0003] In related technologies, the scanning time for each MRI slice has been effectively shortened, greatly improving the speed of MRI imaging. However, the reduced scanning time has impacted the image quality of each acquired MRI slice, showing a negative correlation between scanning time and image quality. Therefore, achieving both high-speed and high-quality MRI imaging is a problem that needs to be solved.

[0004] Application content

[0005] This application aims to at least partially solve one of the technical problems in the aforementioned technologies.

[0006] Therefore, the first aspect of this application proposes a scanning method for magnetic resonance imaging (MRI).

[0007] The second aspect of this application provides a scanning device for magnetic resonance imaging (MRI).

[0008] The third aspect of this application proposes an electronic device.

[0009] The fourth aspect of this application proposes a readable storage medium.

[0010] The fifth aspect of this application proposes a computer program product.

[0011] To achieve the above objectives, the first aspect of this application proposes a magnetic resonance imaging (MRI) scanning method, comprising: acquiring raw K-space data of the current scanning slice within each scanning slice segment, and generating a first slice image of the current scanning slice based on the raw K-space data; determining whether the current scanning slice is a first type of scanning slice, wherein the first type of scanning slice is imaged using an intra-frame imaging mode; in response to the current scanning slice not being a first type of scanning slice, determining the adjacent scanning slice corresponding to the current scanning slice, and acquiring a second slice image of the adjacent scanning slice, wherein the resolution of the second slice image is greater than the resolution of the first slice image; and performing imaging prediction on the current scanning slice using an inter-frame prediction mode based on the first slice image and the second slice image of the adjacent scanning slice, thereby generating a second slice image of the current scanning slice.

[0012] In addition, the MRI scanning method proposed in the first aspect of this application may also have the following technical features:

[0013] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning method further includes: if the current scanning plane is a first type of scanning plane, performing intra-MR imaging on the current scanning plane based on the first plane image to generate a second plane image of the current scanning plane.

[0014] According to one embodiment of this application, before performing imaging prediction on the current scanning layer using an MR inter-frame prediction mode based on the first layer image and the second layer image of the adjacent scanning layer, the method further includes: in response to the current scanning layer being a second type of scanning layer, obtaining the structural similarity between the first layer image and the first layer image and / or the second layer image of the adjacent scanning layer, and determining that the structural similarity is greater than a preset similarity threshold, wherein the second type of scanning layer is imaged using a unidirectional inter-frame prediction mode.

[0015] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning method further includes: if the structural similarity is less than or equal to the preset similarity threshold, modifying the next scanning layer to the current scanning layer, and returning to execute the steps of obtaining the K-space data of the current scanning layer and subsequent steps.

[0016] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning method further includes: if the structural similarity is less than or equal to the preset similarity threshold, saving the K-space data and the first layer image.

[0017] According to one embodiment of this application, after saving the K-space data and the first layer image, the method further includes: when scanning to the scanning position of the current scanning layer in the next scanning layer segment, if the current scanning layer is not the first type of scanning layer of the next scanning layer segment, then the saved first layer image is directly used as the first layer image of the current scanning layer; or, if the current scanning layer is the first type of scanning layer of the next scanning layer segment, the original K-space data is supplemented according to the sampling rate of the first type of scanning layer, and the first layer image of the first type of scanning layer is generated based on the supplemented K-space data.

[0018] According to one embodiment of this application, determining the adjacent scanning layers corresponding to the current scanning layer further includes: obtaining a prediction mode of the current scanning layer, wherein the prediction mode includes unidirectional inter-frame prediction and bidirectional inter-frame prediction; and determining the adjacent scanning layers based on the prediction mode of the current scanning layer.

[0019] According to one embodiment of this application, determining the adjacent scanning layers based on the prediction mode of the current scanning layer includes: if the prediction mode is a unidirectional inter-frame prediction mode, taking the forward prediction scanning layer on which the current layer image is predicted as the adjacent scanning layers; if the prediction mode is the bidirectional inter-frame prediction mode, determining the forward prediction scanning layer and the backward prediction scanning layer on which the current layer image is predicted as the adjacent scanning layers.

[0020] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning method further includes: if the current scanning plane is a set of planes including multiple sub-scanning planes, starting from the first sub-scanning plane of the set of planes, performing MR inter-frame prediction mode on each sub-scanning plane in the set of planes to generate a second plane image corresponding to each sub-scanning plane in the set of planes, until each sub-scanning plane in the set of planes is imaged and the next scanning plane or the next scanning plane segment is entered.

[0021] According to one embodiment of this application, the process of generating a second layer image of the current scanning layer includes: extracting features from an input layer image by a feature extraction layer and outputting a feature image, wherein the input layer image is the first layer image and the second layer image of the adjacent scanning layer when performing MR inter-frame prediction mode, and the first layer image when performing MR intra-frame imaging; generating prediction K-space data from the feature image by a data processing layer, and generating a prediction enhancement layer image of the current scanning layer based on the prediction K-space data; inputting the prediction enhancement layer image into an image enhancement layer, outputting residual information, and generating a second layer image of the current scanning layer based on the residual information and the prediction enhancement layer image.

[0022] According to one embodiment of this application, generating a predictive augmentation layer image of the current scanning layer based on the predicted K-space data includes: replacing the predicted data at the sampling frequency in the predicted K-space data with the original K-space data to generate target K-space data; and generating a predictive augmentation layer image of the current layer based on the target K-space data.

[0023] According to one embodiment of this application, in the magnetic resonance imaging (MRI) scanning method, intra-MR imaging is performed by an intra-MR imaging model, and inter-MR prediction mode is performed by an inter-MR prediction mode model. Both the inter-MR prediction mode model and the intra-MR imaging model include multiple image processing units connected sequentially. Each image processing unit includes a feature extraction layer, a data processing layer, and an image enhancement layer. The feature extraction layer, data processing layer, and image enhancement layer within each image processing unit are connected sequentially, and the output of the image enhancement layer is the input of the feature extraction layer in the next image processing unit.

[0024] To achieve the above objectives, a second aspect of this application provides a magnetic resonance imaging (MRI) scanning apparatus, comprising: a generation module, configured to acquire raw K-space data of the current scanning slice within each scanning slice segment, and generate a first slice image of the current scanning slice based on the raw K-space data; a determination module, configured to determine whether the current scanning slice is a first type of scanning slice, wherein the first type of scanning slice is imaged using an intra-frame imaging mode; an acquisition module, configured to, in response to the current scanning slice not being a first type of scanning slice, determine an adjacent scanning slice corresponding to the current scanning slice, and acquire a second slice image of the adjacent scanning slice, wherein the resolution of the second slice image is greater than the resolution of the first slice image; and an imaging module, configured to perform imaging prediction on the current scanning slice using an inter-frame prediction mode based on the first slice image and the second slice image of the adjacent scanning slice, and generate a second slice image of the current scanning slice.

[0025] In addition, the MRI scanning device proposed in the second aspect of this application may also have the following technical features:

[0026] According to one embodiment of this application, the imaging module is further configured to: if the current scanning plane is the first type of scanning plane, perform MR intra-frame imaging on the current scanning plane based on the first plane image to generate a second plane image of the current scanning plane.

[0027] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning device further includes: a similarity determination module, configured to, in response to the current scanning plane being a second type of scanning plane, acquire the structural similarity between the first plane image and the first plane image and / or the second plane image of the adjacent scanning plane, and determine that the structural similarity is greater than a preset similarity threshold, wherein the second type of scanning plane is imaged using a unidirectional inter-frame prediction mode.

[0028] According to one embodiment of this application, the similarity determination module is further configured to: if the structural similarity is less than or equal to the preset similarity threshold, modify the next scanning layer to the current scanning layer, and return to execute the steps of obtaining the K-space data of the current scanning layer and subsequent steps.

[0029] According to one embodiment of this application, the similarity determination module is further configured to: if the structural similarity is less than or equal to the preset similarity threshold, save the K-space data and the first layer image.

[0030] According to one embodiment of this application, the generation module is further configured to: when scanning to the scanning position of the current scanning layer in the next scanning layer segment, if the current scanning layer is not the first type of scanning layer of the next scanning layer segment, directly use the saved first layer image as the first layer image of the current scanning layer; or, if the current scanning layer is the first type of scanning layer of the next scanning layer segment, supplement the original K-space data according to the sampling rate of the first type of scanning layer, and generate the first layer image of the first type of scanning layer based on the supplemented K-space data.

[0031] According to one embodiment of this application, the acquisition module is further configured to: acquire the prediction mode of the current scanning layer, wherein the prediction mode includes unidirectional inter-frame prediction and bidirectional inter-frame prediction; and determine the adjacent scanning layer based on the prediction mode, using the scanning layer on which the prediction of the current scanning layer depends.

[0032] According to one embodiment of this application, the acquisition module is further configured to: if the prediction mode is a one-way inter-frame prediction mode, take the forward prediction scanning layer on which the current layer image is predicted as the adjacent scanning layer; if the prediction mode is the two-way inter-frame prediction mode, determine the forward prediction scanning layer and the backward prediction scanning layer on which the current layer image is predicted as the adjacent scanning layers.

[0033] According to one embodiment of this application, the magnetic resonance imaging (MRI) scanning device further includes: a scanning module, configured to, if the current scanning plane is a plane set including multiple sub-scanning planes, start from the first sub-scanning plane of the plane set, perform MR inter-frame prediction mode on each sub-scanning plane in the plane set, generate a second plane image corresponding to each sub-scanning plane in the plane set, until the imaging of each sub-scanning plane in the plane set is completed, and enter the next scanning plane or the next scanning plane segment.

[0034] According to one embodiment of this application, the imaging module is further configured to: extract features from the input layer image by a feature extraction layer and output a feature image, wherein the input layer image is the first layer image and the second layer image of the adjacent scanning layer when performing MR inter-frame prediction mode, and the first layer image when performing MR intra-frame imaging; generate prediction K-space data from the feature image by a data processing layer, and generate a prediction enhancement layer image of the current scanning layer based on the prediction K-space data; input the prediction enhancement layer image into an image enhancement layer, output residual information, and generate the second layer image of the current scanning layer based on the residual information and the prediction enhancement layer image.

[0035] According to one embodiment of this application, the imaging module is further configured to: replace the predicted data at the sampling frequency in the predicted K-space data with the original K-space data to generate target K-space data; and generate a predicted augmented layer image of the current layer based on the target K-space data.

[0036] According to one embodiment of this application, in the imaging module, intra-MR imaging is performed by an intra-MR imaging model, and inter-MR prediction mode is performed by an inter-MR prediction mode model. Both the inter-MR prediction mode model and the intra-MR imaging model include multiple image processing units connected sequentially. Each image processing unit includes a feature extraction layer, a data processing layer, and an image enhancement layer. The feature extraction layer, data processing layer, and image enhancement layer within each image processing unit are connected sequentially, and the output of the image enhancement layer is the input of the feature extraction layer in the next image processing unit.

[0037] This application discloses an electronic device in its third aspect, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the magnetic resonance imaging (MRI) scanning method described in the first aspect above.

[0038] The fourth aspect of this application provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to perform a magnetic resonance imaging (MRI) scanning method according to the first aspect.

[0039] The fifth aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the magnetic resonance imaging (MRI) scanning method according to the first aspect.

[0040] The magnetic resonance imaging (MRI) scanning method and apparatus proposed in this application acquire the raw K-space data of the current scanning slice within each scanning slice segment, generate a first-slice image corresponding to the current scanning slice, and then determine whether the current scanning slice belongs to the first type of scanning slice. If it is determined that the current scanning slice does not belong to the first type of scanning slice, a second-slice image corresponding to its adjacent scanning slice is acquired, and a second-slice image corresponding to the current scanning slice is generated based on the first-slice image of the current scanning slice and the second-slice images corresponding to the adjacent scanning slices. In this application, a low-resolution first-slice image corresponding to each scanning slice is acquired first, and then a high-resolution second-slice image is obtained by processing the resolution of the first-slice image, effectively shortening the imaging time. By utilizing the structural correlation between adjacent slice images, the prediction and generation of the first-slice image corresponding to each scanning slice is achieved, effectively shortening the scanning time. Separate scanning imaging of different scanning slices within each scanning slice segment ensures imaging quality, achieving the goal of high-speed, high-quality MRI scanning imaging.

[0041] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0042] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0043] Figure 1 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to an embodiment of this application;

[0044] Figure 2 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0045] Figure 3 This is a schematic flowchart of an MR intra-frame imaging model according to an embodiment of this application;

[0046] Figure 4 This is a schematic diagram of the data processing flow of a DC computing architecture according to an embodiment of this application;

[0047] Figure 5 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0048] Figure 6 This is a flowchart illustrating an embodiment of the MR inter-frame prediction mode model of this application.

[0049] Figure 7 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0050] Figure 8 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0051] Figure 9 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0052] Figure 10 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0053] Figure 11 This is a flowchart illustrating another embodiment of the MR inter-frame prediction mode model of this application;

[0054] Figure 12 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0055] Figure 13 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0056] Figure 14 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application;

[0057] Figure 15 This is a schematic diagram of the structure of a magnetic resonance imaging (MRI) scanning device according to an embodiment of this application;

[0058] Figure 16 This is a schematic diagram of the structure of a magnetic resonance imaging (MRI) scanning device according to another embodiment of this application;

[0059] Figure 17 This is a block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0060] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0061] The following description, in conjunction with the accompanying drawings, describes the magnetic resonance imaging (MRI) scanning method, apparatus, air conditioning system, electronic equipment, and computer-readable storage medium proposed in this application.

[0062] Figure 1 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to an embodiment of this application, as shown below. Figure 1 As shown, the method includes:

[0063] S101, within each scanning slice, acquire the original K-space data of the current scanning slice, and generate the first slice image of the current scanning slice based on the original K-space data.

[0064] Nuclear magnetic resonance imaging (NMRI), also known as spin imaging or magnetic resonance imaging (MRI), utilizes the principle of nuclear magnetic resonance (NMR). Based on the different attenuations of released energy in different structural environments within a substance, and by detecting the emitted electromagnetic waves through an external gradient magnetic field, the location and type of atomic nuclei that make up the object can be determined, and an image of the object's internal structure can be created.

[0065] Generally, MRI is primarily used in medicine for imaging the structure of human tissues. The human body contains numerous atomic nuclei containing an odd number of protons, such as hydrogen nuclei. Protons possess spin motion. By placing these spin protons in a strong external magnetic field, they align in two directions parallel or antiparallel to the magnetic field. Based on this scenario, excitation with radio frequency pulses of a specific frequency causes the nuclei containing the odd number of protons to absorb a certain amount of energy and resonate, a phenomenon known as magnetic resonance. In practice, after the radio frequency pulse is stopped, the nuclei containing the odd number of protons that produced the magnetic resonance phenomenon gradually release the absorbed energy. Consequently, their phase and energy levels can return to their state before being excited by the radio frequency pulse; this process is called relaxation. Furthermore, the duration of the relaxation process is called the relaxation time.

[0066] Different tissue structures in the human body require different relaxation times, and these differences in relaxation times form the basis of magnetic resonance imaging (MRI). Generally, during an MRI scan, the tissue structure to be examined is divided into multiple imaging layers. By imaging each layer separately, a complete image of the tissue structure is generated.

[0067] During MRI scanning, the image data acquired for each scanning layer is stored in the K-space. The original K-space data corresponding to each scanning layer can be obtained from the K-space, which carries part of the image data of the current scanning layer. The first layer image of the current scanning layer is generated based on this part of the data.

[0068] In practice, different sampling frequencies are suitable for different parts of the human body's tissue structure. During MRI scanning, the sampling frequency needs to be adjusted in real time for the tissue structure covered by each scanning layer to obtain high-accuracy layer image data. Optionally, undersampling can be used to acquire image data for the current scanning layer, and the scanning frequency of the current scanning layer can be adjusted in real time based on the actual situation to obtain accurate image data of the current scanning layer and store it in K-space.

[0069] Furthermore, a set image data processing method can be used to transform the original K-space data into a layered image, such as the inverse fourier transform (IFFT).

[0070] S102, determine whether the current scanning plane is a first type of scanning plane, wherein the first type of scanning plane uses intra-frame imaging mode for imaging.

[0071] The first type of scanning plane is the first scanning plane in each scanning plane segment that begins scanning and imaging.

[0072] Since the generation methods of the first-type scanning plane and the first-layer images of non-first-type scanning planes are different in the magnetic resonance MRI scanning imaging method proposed in this application, it is necessary to determine the type when starting scanning imaging for a certain scanning plane.

[0073] Optionally, the position of the current scan layer within its respective scan layer segment can be used to determine whether the current scan layer is a first-type scan layer. For each scan layer segment, the scan layers are numbered sequentially according to their scanning order to obtain the corresponding number for the current scan layer, thus determining its position within the segment. For example, if the scan layers within a scan layer segment are numbered in the order of "1, 2, 3, 4…", and the obtained number is 1 (the initial number of its segment), then the current scan layer is determined to be a first-type scan layer.

[0074] Optionally, it can also be determined whether the current scanning layer is a type I scanning layer by whether a preceding scanning layer exists. If a preceding scanning layer exists, the current scanning layer is determined to be a non-type I scanning layer; if no preceding scanning layer exists, the current scanning layer is determined to be a type I scanning layer.

[0075] Furthermore, for the current scanning layer that belongs to the first type of scanning layer, the K-space image data stored in the original K-space of the current scanning layer is obtained, and then the first layer image of the current scanning layer is generated. The first layer image is input into the image processing model for further resolution processing to obtain high-quality imaging of the current layer.

[0076] Optionally, the image processing model can be a magnetic resonance (MR) intra-frame imaging model. This can be understood as follows: for the first-level image of the first type of scan plane in each scan segment, image processing can be performed on the pixel information within that first-level image to generate a high-resolution image corresponding to that first-level image. Further, the model that processes the intra-frame pixel information of the first-level image of the first type of scan plane to generate high-quality imaging is defined as the MR intra-frame imaging model.

[0077] S103, in response to the current scanning layer not being a first type of scanning layer, the adjacent scanning layer corresponding to the current scanning layer is determined, and the second layer image of the adjacent scanning layer is obtained, wherein the resolution of the second layer image is greater than the resolution of the first layer image.

[0078] Optionally, it can be determined whether the current scanning layer is a non-first-type scanning layer by obtaining its position within its respective scanning layer segment. Using the example above, if each scanning layer in the scanning layer segment is numbered "1, 2, 3, 4..." according to the scanning sequence, then all scanning layers except for the one numbered 1 can be identified as non-first-type scanning layers.

[0079] Optionally, it can be determined whether the current scanning layer is a non-first type scanning layer by whether there is a preceding scanning layer. If there is a preceding scanning layer, the current scanning layer is determined to be a non-first type scanning layer.

[0080] Optionally, the adjacent scanning layers of the current scanning layer can be determined by the scanning layer number, or it can be determined whether there are other scanning layers between the current scanning layer and its corresponding preceding scanning layer. The preceding scanning layers that do not have other scanning layers between them and the current scanning layer are obtained as the adjacent scanning layers of the current scanning layer, and then the second layer image of the adjacent scanning layers is obtained.

[0081] Furthermore, the low-resolution first-level image is input into the feature extraction image processing layer, such as a preset first deep convolutional neural network, to extract and process image features. After outputting the feature image, it is input into a set data processing layer, such as a computing architecture (Data Consistency, DC), to perform K-space data replacement. The layer image obtained after replacement is input into the image enhancement layer, such as a set second deep convolutional neural network, for further processing, thereby obtaining a high-resolution layer image, which is the second-level image corresponding to the first-level image of any scan layer.

[0082] S104, based on the first-level image and the second-level image of the adjacent scanning level, the imaging prediction of the current scanning level is performed using the MR inter-frame prediction mode to generate the second-level image of the current scanning level.

[0083] Generally, magnetic resonance imaging (MRI) can be used to examine human tissue structures. Any human tissue structure is divided into multiple scanning segments. By imaging each scanning segment and integrating the images generated from each segment, a holistic image of the tissue structure is achieved. Since the multiple scanning segments are based on the same tissue structure, there is structural correlation between the scanned objects corresponding to each segment. Furthermore, each scanning slice within each segment also has structural correlation.

[0084] Based on the second-level images of adjacent scanning levels of the current scanning level, the first-level image of the current scanning level can be obtained. Furthermore, based on the second-level images of adjacent scanning levels and the first-level image of the current scanning level, the second-level image of the current scanning level can be obtained.

[0085] Optionally, the first-level image of the current scanning layer and the second-level image of the adjacent scanning layer can be processed together using an MR inter-frame prediction mode model to obtain the second-level image of the current scanning layer.

[0086] This can be understood as follows: when the current scanning plane is not the first type of scanning plane in its scanning plane segment, the pixel information in the first-level image of the current scanning plane, as well as the pixel information in the second-level image of the adjacent scanning plane, can be processed to generate the second-level image corresponding to the predicted first-level image of the current scanning plane. Furthermore, the model that processes the pixel information in the first-level image of a non-first-type scanning plane and the second-level image of its adjacent scanning plane to generate a predicted high-quality image is defined as the MR inter-frame prediction model.

[0087] It should be noted that in the embodiments of this application, each scanning layer generates a low-resolution first layer image, which in turn generates a high-resolution second layer image. It can be understood that the second layer image has the same content as the first layer image, but has a higher resolution and higher quality, and is the high-quality final image corresponding to each scanning layer.

[0088] The magnetic resonance imaging (MRI) scanning method proposed in this application acquires the original K-space data of the current scanning slice within each scanning slice segment, generates a first-slice image corresponding to the current scanning slice, and then determines whether the current scanning slice belongs to the first type of scanning slice. After determining that the current scanning slice does not belong to the first type of scanning slice, it acquires the second-slice image corresponding to its adjacent scanning slice, and generates a second-slice image corresponding to the current scanning slice based on the first-slice image of the current scanning slice and the second-slice images corresponding to the adjacent scanning slice. In this application, a low-resolution first-slice image corresponding to each scanning slice is first acquired, and then a high-resolution second-slice image is obtained by processing the resolution of the first-slice image, effectively shortening the imaging time. By utilizing the structural correlation between adjacent slice images, the prediction and generation of the first-slice image corresponding to each scanning slice is achieved, effectively shortening the scanning time. Separate scanning imaging of different scanning slices within each scanning slice segment ensures imaging quality, achieving the goal of high-speed, high-quality MRI scanning imaging.

[0089] Based on the above embodiments, for the generation of layer images of the first type of scanning layer, it is possible to combine... Figure 2 To understand further, Figure 2 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 2 As shown, the method includes:

[0090] S201, if the current scanning plane is a first type of scanning plane, perform MR intra-frame imaging on the current scanning plane based on the first plane image to generate a second plane image of the current scanning plane.

[0091] Generally, for the first type of scanning plane, since there are no preceding adjacent scanning planes and the corresponding second-level image will be used as the imaging prediction dependency for subsequent scanning planes, it is necessary to ensure the correctness of the second-level image corresponding to the scanning plane that belongs to the first type of scanning plane.

[0092] Image data corresponding to the current scan level can be obtained from the original K-space data to generate the first-level image of the current scan level. Further, a second-level image of the current scan level is generated using an MR intra-frame imaging model.

[0093] like Figure 3As shown, the first-level image is input into a preset first-depth convolutional neural network to extract image features and obtain the feature image of the first-level image. The obtained feature image is input into the first data processing, such as the DC computing architecture shown in the figure, and K-space data replacement is performed on it. The K-space data-replaced level image is then input into a set second-depth convolutional neural network to perform residual processing of the level image and output residual information. The residual information is fused with the resolution-enhanced level image output by DC to generate the second-level image.

[0094] Among them, the operation mode of DC is as follows Figure 4 As shown, the feature image of the first layer image is input to the DC. The DC can perform data transformation on the input feature image, such as converting the feature image to image data using a Fourier transform (FFT). The image data obtained through image transformation processing has the same format as the scanned image data stored in the K-space. Furthermore, a portion of the image data obtained from scanning this layer in the K-space can be replaced in the transformed image data. The K-space stores a portion of the scanned image data belonging to the first layer image; the amount of this portion of image data is determined by the sampling frequency of the first type of scan layer.

[0095] It should be noted that, depending on the MRI scanning device and the resolution requirements of the MRI imaging, the above-mentioned second-level image processing process can be repeated multiple times until the final output image resolution meets the requirements.

[0096] The magnetic resonance imaging (MRI) scanning method proposed in this application enhances the resolution of the first-slice image of the first type of scanning plane using an intra-MR imaging model, thereby ensuring the imaging quality and accuracy of the first type of scanning plane. This effectively guarantees the accuracy of subsequent predictive imaging based on the second-slice image of the first type of scanning plane.

[0097] In the above embodiments, for the acquisition of layer images that are not of the first type of scanning layer, it can be combined with Figure 5 To understand further, Figure 5 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 5 As shown, the method includes:

[0098] S501, in response to the current scanning layer being a second type of scanning layer, obtain the structural similarity between the first layer image and the first layer image and / or the second layer image of the adjacent scanning layer, and determine that the structural similarity is greater than a preset similarity threshold, wherein the second type of scanning layer uses a unidirectional inter-frame prediction mode for imaging.

[0099] In this embodiment of the application, the scanning layers contained in each scanning layer segment can be divided into a first type of scanning layer and a non-first type of scanning layer. The non-first type of scanning layer may include a second type of scanning layer, and the second type of scanning layer has a preceding adjacent scanning layer.

[0100] For the current scanning layer belonging to the second type of scanning layer, the first layer image of the current scanning layer can be obtained by predicting from the second layer images of its preceding adjacent scanning layers based on the structural correlation between the current scanning layer and its preceding adjacent scanning layers. The first layer image of the current scanning layer belonging to the second type of scanning layer can be obtained through a unidirectional inter-frame prediction mode.

[0101] Among them, different scanning layers contained in the same scanning layer segment have structural correlations. In particular, two adjacent scanning layers correspond to adjacent parts of the scanned object. Therefore, two adjacent scanning layers have a stronger correlation, which can be similarity.

[0102] After obtaining the first layer image of the current scanning layer, a similarity threshold can be set. If the structural similarity between the first layer image of the current scanning layer and the first layer image and / or second layer image of the preceding adjacent scanning layer is greater than the preset similarity threshold, it can be determined that the structure of the scanning object corresponding to the current scanning layer is similar to that of the preceding scanning layer, and no abnormal changes have occurred. Therefore, the obtained first layer image of the current scanning layer is reasonable and usable.

[0103] Furthermore, a high-resolution, high-quality second-layer image can be obtained based on the first-layer image of the current scanning layer. Optionally, resolution can be adjusted using an image processing model, such as an MR inter-frame prediction mode model. Figure 6 As shown.

[0104] The second-level image of the previous scanning layer and the first-level image of the current scanning layer can be simultaneously input into a set first-depth convolutional neural network to extract image features, obtain the feature image of the first-level image, and then input the obtained feature image into the DC computing architecture to replace the K-space data. For the specific method of replacing the K-space data of the DC computing architecture, please refer to the above-mentioned detailed content, which will not be repeated here.

[0105] Furthermore, the resolution-enhanced image output by the DC is input into a predefined second deep convolutional neural network for residual processing, and the residual information is output. Based on the residual information output by the second deep convolutional neural network and the resolution-enhanced image output by the DC, a high-resolution second-layer image is generated.

[0106] It should be noted that the processing of the first-level image can be carried out in multiple cycles according to different needs, so as to achieve the set standard of the layer image.

[0107] The MRI scanning method proposed in this application obtains the first-level image of the current scanning level based on the prediction of the second-level images of the preceding adjacent scanning levels using structural similarity, and generates a higher-resolution second-level image of the current scanning level. In this application, obtaining the first-level image of the current level through prediction effectively shortens the imaging time. Furthermore, the simultaneous image processing of the second-level images of the preceding adjacent scanning levels and the first-level image of the current scanning level using an inter-frame prediction model ensures both imaging accuracy and quality.

[0108] Based on the above embodiments, when the structural similarity is less than or equal to a preset similarity threshold, it can be combined with Figure 7 , Figure 7 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 7 As shown, the method includes:

[0109] S701, if the structural similarity is less than or equal to the preset similarity threshold, modify the next scanning layer to the current scanning layer, and return to execute the process of obtaining the K-space data of the current scanning layer and subsequent steps.

[0110] Based on the above embodiments, when the structural similarity between the first layer image of the current scanning layer and the first layer image and / or the second layer image of the previous scanning layer is less than or equal to a preset similarity threshold, it indicates that the structural similarity between the current scanning layer and the previous scanning layer is low, and it can be determined that the structure of the scanning object corresponding to the current scanning layer has changed significantly compared with the scanning object corresponding to the previous scanning layer.

[0111] Since the overall imaging of an MRI scan is obtained by fusing images generated from each scan slice, and the final imaging of each scan slice is obtained by fusing second-level images from all scan slices within that slice, if the structural similarity between the current scan slice and the corresponding slice images of the preceding adjacent scan slices is less than a set similarity threshold, and if the first-level image of the current scan slice is constructed based on the second-level images of the preceding adjacent scan slices, the first-level image of the current scan slice will not match the actual situation of the scanned object, resulting in inaccurate imaging results. Furthermore, if the first-level image of the next scan slice is constructed based on the imaging results of the current scan slice, the imaging results of the next scan slice will also be inaccurate, thus affecting the accuracy of the final imaging result of the scan slice to which the current scan slice belongs.

[0112] Furthermore, since the current scanning plane is close to the scanning object corresponding to the previous scanning plane, a new scanning plane segment needs to be constructed for the part corresponding to the changed scanning object in order to perform complete imaging.

[0113] Furthermore, the current scanning plane can serve as the scanning plane for the next scanning plane segment, allowing the MRI scanning imaging process to be adjusted in a timely manner according to different actual situations. Areas in the scanned object that have undergone structural changes can be focused on imaging to ensure the accuracy and usability of the imaging results.

[0114] Optionally, the current scanning plane can be used as the first scanning plane of the next scanning plane segment, i.e., as a first-type scanning plane, or the current scanning plane can be used as a scanning plane in another position in the next scanning plane segment that is not the first, i.e., as a non-first-type scanning plane for scanning imaging.

[0115] In this scenario, if the current scanning plane is taken as the first scanning plane in the next scanning plane segment, then this scanning plane belongs to the first type of scanning plane in the next scanning plane segment. By acquiring its corresponding K-space data, a first-level image can be generated, thereby enabling subsequent related imaging operations to acquire the corresponding second-level image. If the current scanning plane is taken as a scanning plane that is not the first in the next scanning plane segment, then this scanning plane belongs to a non-first-level scanning plane in the next scanning plane segment. When it is the turn of this scanning plane to perform imaging in the next scanning plane segment, the relevant layer image acquisition operations can be performed.

[0116] The magnetic resonance imaging (MRI) scanning method proposed in this application addresses the situation where the first layer image of the current scanning plane is less than or equal to a preset similarity threshold set between it and the second layer image of the previous scanning plane. It proposes a corresponding processing method that can make timely adjustments according to changes in the actual situation during the MRI scanning process, enabling focused imaging of the changed scanning areas, improving the accuracy of MRI scanning imaging, and enhancing the usability of the imaging results.

[0117] Furthermore, it can be combined with Figure 8 , Figure 8 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 8 As shown, the method includes:

[0118] S801, if the structural similarity is less than or equal to the preset similarity threshold, save the K-space data and the first-level image.

[0119] Generally, to reduce the time required to acquire image data during MRI scanning and / or prediction, the acquired scan image data is stored in the K-space. Similarly, the predicted image data for the first slice of a specific scan is also stored. In subsequent acquisitions of slice images through scanning and / or prediction, if image data for the same scan slice is being acquired, the previously saved data can be directly retrieved, effectively shortening the data acquisition time.

[0120] In this embodiment, when the similarity between the second-level image of the currently adjacent scanning layer and the first-level image of the currently scanning layer is less than or equal to a preset similarity threshold, it can be determined that an anomaly has occurred in the prediction of the first-level image of the currently scanning layer, such as low clarity of the first-level image. In this scenario, the image data obtained from the current scanning layer is stored in the K-space, and the image predicted based on the second-level image of the preceding adjacent layer is stored in a set location.

[0121] Furthermore, the current scan plane can be removed from its current scan plane segment and added to the next scan plane segment.

[0122] As one possible implementation, the current scanning plane can be one of the scanning planes in the next scanning plane segment. When the next scanning plane segment scans to the scanning position of the current scanning plane, if the current scanning plane is not the first type of scanning plane in the next scanning plane segment, the saved first-plane image is directly used as the first-plane image of the current scanning plane.

[0123] In this embodiment of the application, if the current scanning layer is not a first type of scanning layer in the next scanning layer segment, then when imaging the scanning layer, the stored historical scanning data can be directly used to generate the first layer image corresponding to the scanning layer.

[0124] Generally, for the first-level image of a scan layer that does not belong to the first type of scan layer, prediction is made based on the second-level image of its preceding adjacent scan layer, and subsequent imaging steps are performed based on the similarity between the two. Therefore, if historical image data exists for the current scan layer, the historically stored layer image data can be directly called, and the first-level image of the current scan layer can be generated based on the historical layer image data. By calling historical data, the generation time of the first-level image can be effectively shortened, while ensuring the accuracy of the first-level image.

[0125] As another possible implementation, the current scanning plane can be used as the initial scanning plane for the next scanning plane. That is, if the current scanning plane is the first type of scanning plane of the next scanning plane segment, the original K-space data is supplemented according to the sampling rate of the first type of scanning plane, and the first layer image of the first type of scanning plane is generated based on the supplemented K-space data.

[0126] In this embodiment of the application, the current scanning plane can be used as the first type of scanning plane in the next scanning plane segment.

[0127] In implementation, the K-space stores image data of scanned layers that have already been sampled. The first-layer image of a scanned layer belonging to the first type of scanned layer can be generated by acquiring the image data from the K-space. Furthermore, when the scanned layer belongs to the first type of scanned layer, the generation of its first-layer image requires complete sampled image data of that scanned layer.

[0128] Generally, due to the sampling frequency, the historical image data stored in the K-space is only a portion of the image data obtained from the sampling of that scanning plane, and its data volume is determined by the sampling frequency of that scanning plane. When the scanning plane is a first-type scanning plane, it is necessary to supplement the image data of the unsampled parts of that scanning plane to obtain the complete image data of that scanning plane.

[0129] It should be noted that, in order to ensure that the format of the image data obtained by supplementary sampling is consistent with the historical data stored in the K space, the sampling frequency can be adjusted to be consistent with the sampling rate of the first type of scanning layer when supplementing the image data.

[0130] Furthermore, after the supplementary sampling is completed, the K-space data can store the complete image data of the scanned layer, and then the first layer image of the scanned layer can be generated based on the K-space data.

[0131] The magnetic resonance imaging (MRI) scanning method proposed in this application addresses the issue of the current scanning layer being located at different positions in the next scanning layer when the similarity between the first layer image of the scanning layer and the second layer image of the adjacent scanning layer is lower than a preset similarity threshold. This effectively shortens the MRI scanning imaging time and ensures the accuracy of the imaging.

[0132] Figure 9 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 9 As shown, the method includes:

[0133] S901, Obtain the prediction mode of the current scanning layer, wherein the prediction mode includes unidirectional inter-frame prediction and bidirectional inter-frame prediction.

[0134] One-way inter-frame prediction is a prediction method that uses the structural similarity between the current scan layer and its preceding adjacent scan layers to predict the second layer image based on the preceding adjacent scan layers, and then obtains the first layer image of the current scan layer.

[0135] Bidirectional inter-frame prediction is a prediction method that uses the second-level image of the current scanning layer and the second-level image of its preceding adjacent scanning layer to predict the layer image between the current scanning layer and its preceding adjacent scanning layer, thereby obtaining the layer image set between the two scanning layers.

[0136] In this embodiment of the application, the required prediction mode can be determined according to the type of layer image to be acquired.

[0137] In general, in order to improve the imaging quality of each scanning segment, as many slice images as possible are generated within each scanning segment. By fusing the slice images corresponding to multiple scanning segments, the corresponding scanning segment can output high-quality MRI scan images.

[0138] Alternatively, a one-way inter-frame prediction method can be used to predict the first layer image of the current scanning layer based on the second layer image of the preceding adjacent scanning layer.

[0139] Optionally, a bidirectional inter-frame prediction method can be used to generate a set of layer images between two adjacent scanning layers based on the second layer images of the previous adjacent scanning layers and the second layer image of the current scanning layer.

[0140] like Figure 10 As shown, let I1 be the second layer image of the scanning layer belonging to the first type of scanning layer in the current scanning layer segment. Based on I1, unidirectional inter-frame prediction can generate the layer image P2 of the next scanning layer. Based on I1 and P2, bidirectional inter-frame prediction can generate a layer image set between the scanning layer corresponding to I1 and the scanning layer corresponding to P2. This layer image set includes layer image B1 and layer image B2.

[0141] Furthermore, based on P2, unidirectional inter-frame prediction is performed to generate the next scan layer image P3. Based on P2 and P3, bidirectional inter-frame prediction is performed to obtain the layer image set containing layer image B3 and layer image B4 between two adjacent scan layers corresponding to P2 and P3.

[0142] It should be noted that B1, B2, B3, and B4 are just examples. In practice, there can be multiple layer images between two adjacent scanning layers, which is not limited here.

[0143] S902, based on the prediction pattern, determines the adjacent scanning layers by referring to the scanning layers on which the prediction of the current scanning layer is based.

[0144] One-way inter-frame prediction requires layer images based on the preceding adjacent scan layers; two-way inter-frame prediction requires layer images based on the preceding layer images of the required layer images and the layer images of the next two adjacent scan layers.

[0145] Optionally, if the prediction mode is a one-way inter-frame prediction mode, the forward-predicted scanning layer on which the current scanning layer's layer image prediction depends is taken as the adjacent scanning layer.

[0146] Still as Figure 10 As shown, P2 is generated based on a unidirectional prediction of I1, and P3 is also generated based on a unidirectional prediction of P2. That is, I1 is the layer image of the forward-predicted scanning layer that P2 relies on when performing unidirectional inter-frame prediction. Therefore, the scanning layer corresponding to I1 can be determined as the adjacent scanning layer of the scanning layer corresponding to P2. Similarly, since P2 is the layer image of the forward-predicted scanning layer that P3 relies on when performing unidirectional inter-frame prediction, the scanning layer corresponding to P2 can be determined as the adjacent scanning layer of the scanning layer corresponding to P3.

[0147] Optionally, if the prediction mode is a bidirectional inter-frame prediction mode, the forward prediction scan layer and the backward prediction scan layer on which the current layer image is predicted are determined as adjacent scan layers.

[0148] like Figure 10 As shown, B1 and B2 are generated based on bidirectional inter-frame prediction using I1 and P2. According to the order shown in the figure, I1 is the layer image of the forward-predicted scanning layer on which B1 and B2 are generated, and P2 is the layer image of the backward-predicted scanning layer on which B1 and B2 are generated. Therefore, the scanning layer corresponding to I1 and the scanning layer corresponding to P2 can be determined as the adjacent scanning layers of B1 and B2.

[0149] Similarly, B3 and B4 are generated based on bidirectional inter-frame prediction of P2 and P3. According to the order shown in the figure, P2 is the layer image of the forward-predicted scanning layer on which B3 and B4 are generated, and P3 is the layer image of the backward-predicted scanning layer on which B3 and B4 are generated. Therefore, the scanning layer corresponding to P2 and the scanning layer corresponding to P3 can be determined as the adjacent scanning layers of B3 and B4.

[0150] It should be noted that subsequent bidirectional inter-frame prediction will only be performed if the structural similarity between the layer image of the current scan layer and the layer image of the previously adjacent scan layer is greater than a preset similarity threshold. This can be understood as... Figure 10In the process, only when the similarity between P2 and I1 is greater than a preset threshold will the next step of bidirectional inter-frame prediction based on I1 and P2 be performed to generate B1 and B2. Alternatively, only when the similarity between P3 and P2 is greater than a preset threshold will the next step of bidirectional inter-frame prediction based on P2 and P3 be performed to generate B3 and B4.

[0151] Optionally, an MR inter-frame prediction mode network model can be used to generate high-resolution second-level images corresponding to the scan layers of type B1, B2, B3, and B4.

[0152] like Figure 11 As shown, to more accurately improve resolution, the low-resolution first-layer image obtained from bidirectional inter-frame prediction and the second-layer images of its two adjacent scan layers can be simultaneously input into a set first-depth convolutional neural network for image feature extraction, generating a feature image and outputting it. Further, the feature image output from the first-depth convolutional neural network is input into a DC computing architecture for K-space data replacement, and the layer image after K-space data replacement is input into a set second-depth convolutional neural network for residual processing, outputting residual information. The residual information output from the second-depth convolutional neural network is then fused with the resolution-enhanced layer image output from the DC architecture to generate the corresponding second-layer image.

[0153] The magnetic resonance imaging (MRI) scanning method proposed in this application obtains the first-level image of the scanning layer through different inter-frame prediction methods according to the requirements of different scanning layer images, which effectively shortens the MRI scanning imaging time and ensures the imaging quality.

[0154] Furthermore, such as Figure 12 As shown, Figure 12 This is a flowchart illustrating a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application. The method includes:

[0155] S1201, if the current scanning plane is a set of planes including multiple sub-scanning planes, starting from the first sub-scanning plane in the set, perform MR inter-frame prediction mode on each sub-scanning plane in the set to generate the second plane image corresponding to each sub-scanning plane in the set, until the imaging of each sub-scanning plane in the set is completed, and enter the next scanning plane or the next scanning plane segment.

[0156] In this embodiment of the application, the current scanning layer may not contain any scanning sub-layers and may be a single scanning layer, or it may contain multiple scanning sub-layers and be generated by combining multiple sub-layers.

[0157] When the current scanning plane contains sub-scanning planes, all sub-scanning planes within the current scanning plane need to be imaged sequentially according to the scanning order. For the current sub-scanning plane, unidirectional inter-frame prediction is performed based on the plane images of its preceding adjacent sub-scanning planes to obtain the first plane image of the current sub-scanning plane. Furthermore, the second plane image of the sub-scanning plane is generated through the MR inter-frame prediction mode. The same operation is performed on all sub-scanning planes until the second plane images of all sub-scanning planes are obtained, at which point the scanning imaging of the current scanning plane can be completed.

[0158] If the current scanning plane is the last scanning plane in its corresponding scanning plane segment, then after completing the imaging of the current scanning plane, you can proceed to the next scanning plane segment to begin imaging of the first scanning plane in the next scanning plane segment and subsequent operations. If the current scanning plane is not the last one in its corresponding scanning plane segment, then after completing the imaging of the current scanning plane, you can directly proceed to the next scanning plane and obtain the first image of the next scanning plane.

[0159] The MRI scanning method proposed in this application requires generating layer images sequentially when the scanning slice contains multiple scanning slices or the scanning slice contains multiple sub-scanning slices, so that the final output layer image can achieve high-quality presentation.

[0160] In the above embodiments, the acquisition of the second-layer image of the current scanning layer can be combined with Figure 13 To understand further, Figure 13 This is a schematic flowchart of a magnetic resonance imaging (MRI) scanning method according to another embodiment of this application, as shown below. Figure 13 As shown, the method includes:

[0161] S1301, the feature extraction layer extracts features from the input layer image and outputs a feature image. The input layer image is the first layer image and the second layer image of the adjacent scanning layer when performing MR inter-frame prediction mode, and the first layer image when performing MR intra-frame imaging.

[0162] Generally, the second-level image is obtained by processing the resolution of the first-level image. In this embodiment, the first-level image can be input into the feature extraction layer, such as a set first deep convolutional neural network, to extract features and generate corresponding feature images.

[0163] The feature extraction layer has different feature extraction processes for the first-level images of different types of scanning layers, thereby realizing the extraction of image features from the first-level images of different types of scanning layers.

[0164] For example, for the first scanning layer belonging to the first type of scanning layer in each scanning layer segment, its first layer image can be generated based on the image data stored in K-space, and the image resolution can be enhanced through the MR intra-frame imaging model to generate the second layer image. In the MR intra-frame imaging model, the feature extraction layer can be a set first deep convolutional neural network to achieve separate processing of the first layer image.

[0165] For example, if the current scanning plane is not a first-class scanning plane, the first-level image of the current scanning plane can be obtained by unidirectional inter-frame prediction based on the second-level image of the preceding adjacent scanning plane, or by bidirectional inter-frame prediction based on the second-level image of the preceding adjacent scanning plane and the second-level image of the reverse adjacent scanning plane. The first-level image of the current scanning plane obtained by unidirectional inter-frame prediction and / or bidirectional inter-frame prediction can be used to enhance the image resolution through the MR inter-frame prediction mode model, thereby generating the second-level image.

[0166] Specifically, for the first layer image of the scanning layer obtained by unidirectional inter-frame prediction, the second layer image of the adjacent scanning layer and the first layer image of the current scanning layer can be simultaneously input into the MR inter-frame prediction mode model. The feature extraction layer in the MR inter-frame prediction mode model can simultaneously extract and fuse features from the two layer images to generate the feature image corresponding to the first layer image.

[0167] Specifically, for the first layer image of the scanning layer obtained by bidirectional inter-frame prediction, the second layer image of the forward adjacent scanning layer, the second layer image of the reverse adjacent scanning layer, and the first layer image of the current scanning layer can be simultaneously input into the MR inter-frame prediction mode model. The feature extraction layer in the MR inter-frame prediction mode model can simultaneously extract and fuse features from the three layer images to generate the feature image corresponding to the first layer image.

[0168] S1302, the data processing layer generates predicted K-space data from the feature image, and based on the predicted K-space data, generates a predicted enhancement layer image for the current scanning layer.

[0169] Generally, both MR intra-frame imaging models and MR inter-frame prediction model models have a data processing layer. The input of the data processing layer is the output of the feature extraction layer. Through the data processing layer, K-space data replacement can be achieved.

[0170] The data processing layer can be a DC computing architecture. In DC, the corresponding image data can be obtained by Fourier transforming the feature image, and the transformed image data is compared with the image data stored in K space. The part that can be covered by the image data stored in K space is replaced, and then the replaced image data is transformed into an enhanced layer image by inverse Fourier transform.

[0171] This can be understood as replacing the predicted data at the sampling frequency in the predicted K-space data with the original K-space data to generate the target K-space data.

[0172] Generally, K-space image data is acquired based on sampling, and only a portion of the image data is stored in K-space based on the sampling frequency. Meanwhile, the first-level image of the current scanning level, generated by predicting the second-level image from adjacent scanning levels, carries all the predicted image data from the current scanning level.

[0173] Therefore, in order to improve the accuracy of the first-level image of the current scanning layer, the image data obtained by prediction can be replaced with the sampled image data stored in the K space. The replaced data is the image data in all the predicted image data that is in the same position as the image data stored in the K space, thereby generating the target K space data.

[0174] like Figure 14 As shown in (a), the gray area can be understood as the image data of the first layer of the current scanned layer, which is sampled and stored in the K-space. Since there is a sampling frequency during scanning, only a portion of the image data of the current scanned layer is obtained. The white area represents the blank area where no image data was obtained through scanning sampling. The black area represents the image data. The image data of the first layer of the current scanned layer obtained through prediction can be as follows... Figure 14 As shown in (b). Further, by replacing the image data of the first layer image of the current scanning layer in 14(b) which is in the same position as in 14(a), the target K-space data of the first layer image of the current scanning layer can be generated, as shown in 14(c).

[0175] Furthermore, a predicted augmented layer image for the current layer is generated based on the target K-space data.

[0176] In this embodiment of the application, after K-space data replacement, the accuracy of the obtained target K-space data is higher. Based on the higher accuracy image data, a layer image with a higher resolution than the input feature image can be generated, which is the prediction enhancement layer image of the current layer.

[0177] S1303, the predicted enhancement layer image is input into the image enhancement layer, residual information is output, and a second layer image of the current scan layer is generated based on the residual information and the predicted enhancement layer image.

[0178] After obtaining the predicted enhancement layer image, it can be input into the image enhancement layer, which can be a second deep neural network. The second deep neural network processes the residual of the input image and outputs the enhancement information of the residual. Furthermore, the enhancement information of the output residual is fused with the layer image input to the second deep neural network to generate the second layer image of the current scan layer.

[0179] It should be noted that intra-MR imaging is performed by the intra-MR imaging model, and inter-MR prediction mode is performed by the inter-MR prediction mode model. Both the inter-MR prediction mode model and the intra-MR imaging model include multiple image processing units, which are connected sequentially. Each image processing unit includes a feature extraction layer, a data processing layer, and an image enhancement layer. The feature extraction layer, data processing layer, and image enhancement layer within each image processing unit are connected sequentially, and the output of the image enhancement layer is the input of the feature extraction layer in the next image processing unit.

[0180] The number of image processing units included in the MR inter-frame prediction mode model and the MR intra-frame imaging model can be set manually or determined based on the resolution requirements of the second-level image at the current level, or based on the relevant parameters of the MRI scanning equipment.

[0181] It should be noted that, in addition to the methods mentioned in the above embodiments, the method for performing resolution enhancement processing on the first layer image of the current scanning layer to generate the second layer image of the current scanning layer also includes other image processing methods that can achieve resolution enhancement of the first layer image of the current scanning layer, which are not limited here.

[0182] The magnetic resonance imaging (MRI) scanning method proposed in this application illustrates the process of generating a second-level image, which enables a low-resolution first-level image to generate a higher-resolution second-level image through a corresponding imaging model, effectively ensuring the imaging quality of the MRI scan.

[0183] Corresponding to the magnetic resonance imaging (MRI) scanning methods proposed in the above embodiments, an embodiment of this application also proposes a magnetic resonance imaging (MRI) scanning device. Since the magnetic resonance imaging (MRI) scanning device proposed in this application corresponds to the magnetic resonance imaging (MRI) scanning methods proposed in the above embodiments, the implementation methods of the above magnetic resonance imaging (MRI) scanning methods are also applicable to the magnetic resonance imaging (MRI) scanning device proposed in this application, and will not be described in detail in the following embodiments.

[0184] Figure 15 This is a schematic diagram of the structure of a magnetic resonance imaging (MRI) scanning device according to an embodiment of this application, as shown below. Figure 15 As shown, the MRI scanning device 100 includes a generation module 11, a judgment module 12, an acquisition module 13, and an imaging module 14, wherein:

[0185] The generation module 11 is used to acquire the original K-space data of the current scanning layer in each scanning layer segment, and generate the first layer image of the current scanning layer based on the original K-space data.

[0186] The determination module 12 is used to determine whether the current scanning plane is a first type of scanning plane, wherein the first type of scanning plane uses an intra-frame imaging mode for imaging;

[0187] The acquisition module 13 is used to determine the adjacent scanning layer corresponding to the current scanning layer in response to the current scanning layer not being the first type of scanning layer, and to acquire the second layer image of the adjacent scanning layer, wherein the resolution of the second layer image is greater than the resolution of the first layer image.

[0188] The imaging module 14 is used to perform imaging prediction on the current scanning layer based on the first layer image and the second layer image of the adjacent scanning layer, using the MR inter-frame prediction mode, to generate the second layer image of the current scanning layer.

[0189] Figure 16 This is a schematic diagram of the structure of a magnetic resonance imaging (MRI) scanning device according to an embodiment of this application, as shown below. Figure 16 As shown, the MRI scanning device 200 includes a generation module 21, a judgment module 22, an acquisition module 23, an imaging module 24, a similarity judgment module 25, and a scanning module 26, wherein:

[0190] It should be noted that the generation module 21, judgment module 22, acquisition module 23, and imaging module 24 have the same structure and function as the generation module 11, judgment module 12, acquisition module 13, and imaging module 14.

[0191] In this embodiment of the application, the imaging module 24 is further configured to: if the current scanning layer is a first type of scanning layer, perform MR intra-frame imaging on the current scanning layer based on the first layer image to generate a second layer image of the current scanning layer.

[0192] In this embodiment of the application, the magnetic resonance imaging (MRI) scanning device 200 further includes:

[0193] The similarity determination module 25 is used to obtain the structural similarity between the first layer image and the first layer image and / or the second layer image of the adjacent scanning layer in response to the current scanning layer being the second type of scanning layer, and to determine that the structural similarity is greater than a preset similarity threshold. The second type of scanning layer uses a unidirectional inter-frame prediction mode for imaging.

[0194] In this embodiment of the application, the similarity determination module 25 is further configured to: if the structural similarity is less than or equal to a preset similarity threshold, modify the next scanning layer to the current scanning layer, and return to execute the acquisition of K-space data of the current scanning layer and subsequent steps.

[0195] In this embodiment of the application, the similarity determination module 25 is further configured to: if the structural similarity is less than or equal to a preset similarity threshold, save the K-space data and the first layer image.

[0196] In this embodiment of the application, the generation module 21 is further configured to: when scanning to the scanning position of the current scanning layer in the next scanning layer segment, if the current scanning layer is not the first type of scanning layer of the next scanning layer segment, directly use the saved first layer image as the first layer image of the current scanning layer; or, if the current scanning layer is the first type of scanning layer of the next scanning layer segment, supplement the original K-space data according to the sampling rate of the first type of scanning layer, and generate the first layer image of the first type of scanning layer based on the supplemented K-space data.

[0197] In this embodiment of the application, the acquisition module 23 is further configured to: acquire the prediction mode of the current scanning layer, wherein the prediction mode includes unidirectional inter-frame prediction and bidirectional inter-frame prediction; and determine the adjacent scanning layers based on the prediction mode of the current scanning layer.

[0198] In this embodiment of the application, the acquisition module 23 is further configured to: if the prediction mode is a one-way inter-frame prediction mode, take the forward prediction scanning layer on which the current layer image is predicted as an adjacent scanning layer; if the prediction mode is a two-way inter-frame prediction mode, determine the forward prediction scanning layer and the reverse prediction scanning layer on which the current layer image is predicted as adjacent scanning layers.

[0199] In this embodiment of the application, the magnetic resonance imaging (MRI) scanning device 200 further includes:

[0200] The scanning module 26 is used to perform MR inter-frame prediction mode on each sub-scanning layer in the layer set starting from the first sub-scanning layer in the layer set if the current scanning layer is a layer set including multiple sub-scanning layers, and generate a second layer image corresponding to each sub-scanning layer in the layer set until the imaging of each sub-scanning layer in the layer set is completed, and then enter the next scanning layer or the next scanning layer segment.

[0201] In this embodiment, the imaging module 24 is further configured to: extract features from the input layer image by the feature extraction layer and output a feature image, wherein the input layer image is a first layer image and a second layer image of an adjacent scanning layer when performing MR inter-frame prediction mode, and a first layer image when performing MR intra-frame imaging; generate prediction K-space data from the feature image by the data processing layer, and generate a prediction enhancement layer image of the current scanning layer based on the prediction K-space data; input the prediction enhancement layer image into the image enhancement layer, output residual information, and generate a second layer image of the current scanning layer based on the residual information and the prediction enhancement layer image.

[0202] In this embodiment of the application, the imaging module 24 is further configured to: replace the predicted data at the sampling frequency in the predicted K-space data with the original K-space data to generate target K-space data; and generate a predicted enhancement layer image of the current layer based on the target K-space data.

[0203] In this embodiment of the application, the imaging module 24 is further configured to: perform intra-MR imaging by an intra-MR imaging model and perform inter-MR prediction mode by an inter-MR prediction mode model, wherein both the inter-MR prediction mode model and the intra-MR imaging model include multiple image processing units connected in sequence, wherein each image processing unit includes a feature extraction layer, a data processing layer and an image enhancement layer, wherein the feature extraction layer, the data processing layer and the image enhancement layer in each image processing unit are connected in sequence, and the output of the image enhancement layer is the input of the feature extraction layer in the next image processing unit.

[0204] The magnetic resonance imaging (MRI) scanning device proposed in this application acquires the raw K-space data of the current scanning slice within each scanning slice segment, generates a first-slice image corresponding to the current scanning slice, and then determines whether the current scanning slice belongs to the first type of scanning slice. After determining that the current scanning slice does not belong to the first type of scanning slice, it acquires the second-slice image corresponding to its adjacent scanning slice, and generates a second-slice image corresponding to the current scanning slice based on the first-slice image of the current scanning slice and the second-slice images corresponding to the adjacent scanning slice. In this application, a low-resolution first-slice image corresponding to each scanning slice is acquired first, and then a high-resolution second-slice image is obtained by processing the resolution of the first-slice image, effectively shortening the imaging time. By utilizing the structural correlation between adjacent slice images, the predictive generation of the first-slice image corresponding to each scanning slice is achieved, effectively shortening the scanning time. Separate scanning imaging of different scanning slices within each scanning slice segment ensures imaging quality, achieving the goal of high-speed, high-quality MRI scanning imaging.

[0205] To achieve the above embodiments, this application also proposes an electronic device, a computer-readable storage medium, and a computer program product.

[0206] Figure 17 A schematic block diagram of an example electronic device 1700 that can be used to implement embodiments of this application is shown. The electronic device 1700 includes a memory 171 and a processor 172. The processor 172 reads executable program code stored in the memory 171 to run a program corresponding to the executable program code, thereby implementing the aforementioned magnetic resonance imaging (MRI) scanning method.

[0207] The electronic device in this application embodiment executes a computer program stored in the memory 171 through the processor 172 to obtain the original K-space data of the current scanning layer within each scanning slice segment, generate a first slice image corresponding to the current scanning layer, and then determine whether the current scanning layer is a first type of scanning layer. After determining that the current scanning layer does not belong to the first type of scanning layer, it obtains the second slice image corresponding to its adjacent scanning layer, and generates a second slice image corresponding to the current scanning layer based on the first slice image of the current scanning layer and the second slice images corresponding to the adjacent scanning layers. In this application, a low-resolution first slice image corresponding to each scanning layer is obtained first, and then a high-resolution second slice image is obtained by processing the resolution of the first slice image, which effectively shortens the imaging time. By utilizing the structural correlation between adjacent slice images, the prediction and generation of the first slice image corresponding to each scanning layer is realized, which effectively shortens the scanning time. Separate scanning imaging is performed on different scanning layers in each scanning slice segment, ensuring imaging quality and achieving the purpose of high-speed and high-quality magnetic resonance MRI scanning imaging.

[0208] This application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described magnetic resonance imaging (MRI) scanning method.

[0209] The computer-readable storage medium of this application, by storing a computer program and having it executed by a processor, acquires the original K-space data of the current scanning layer within each scanning slice segment, generates a first-slice image corresponding to the current scanning layer, and then determines whether the current scanning layer is a first-type scanning layer. After determining that the current scanning layer does not belong to the first-type scanning layer, it acquires the second-slice image corresponding to its adjacent scanning layer, and generates a second-slice image corresponding to the current scanning layer based on the first-slice image of the current scanning layer and the second-slice images corresponding to the adjacent scanning layers. In this application, a low-resolution first-slice image corresponding to each scanning layer is acquired first, and then a high-resolution second-slice image is obtained by processing the resolution of the first-slice image, effectively shortening the imaging time. By utilizing the structural correlation between adjacent slice images, the prediction and generation of the first-slice image corresponding to each scanning layer is achieved, effectively shortening the scanning time. Separate scanning imaging of different scanning layers within each scanning slice segment ensures imaging quality, achieving the goal of high-speed, high-quality magnetic resonance MRI scanning imaging.

[0210] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0211] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0212] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0213] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0214] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0215] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method of scanning for magnetic resonance imaging, MRI, characterized by, include: Within each scanning plane segment, the original K-space data of the current scanning plane is acquired, and a first-plane image of the current scanning plane is generated based on the original K-space data. Determine whether the current scanning layer is a first type of scanning layer, wherein the first type of scanning layer uses an intra-frame imaging mode for imaging; In response to the fact that the current scanning layer is not a first type of scanning layer, the adjacent scanning layer corresponding to the current scanning layer is determined, and the second layer image of the adjacent scanning layer is obtained, wherein the resolution of the second layer image is greater than the resolution of the first layer image. Based on the first layer image and the second layer image of the adjacent scanning layer, the current scanning layer is imaged and predicted using the MR inter-frame prediction mode to generate the second layer image of the current scanning layer. If the current scanning plane is the first type of scanning plane, MR intra-frame imaging is performed on the current scanning plane based on the first plane image to generate a second plane image of the current scanning plane.

2. The method according to claim 1, characterized in that, Before performing imaging prediction on the current scanning layer using the MR inter-frame prediction mode based on the first layer image and the second layer image of the adjacent scanning layers, the method further includes: In response to the current scanning layer being a second type of scanning layer, the structural similarity between the first layer image corresponding to the current scanning layer and the first layer image and / or the second layer image corresponding to the adjacent scanning layer is obtained, and it is determined that the structural similarity is greater than a preset similarity threshold, wherein the second type of scanning layer is imaged using a unidirectional inter-frame prediction mode.

3. The method according to claim 2, characterized in that, Also includes: If the structural similarity is less than or equal to the preset similarity threshold, the next scanning layer is modified to the current scanning layer, and the process of obtaining the K-space data of the current scanning layer and subsequent steps is returned.

4. The method according to claim 3, characterized in that, Also includes: If the structural similarity is less than or equal to the preset similarity threshold, the K-space data and the first layer image are saved.

5. The method according to claim 4, characterized in that, After saving the K-space data and the first layer image, the method further includes: When the next scanning layer reaches the scanning position of the current scanning layer, if the current scanning layer is not the first type of scanning layer of the next scanning layer, then the saved first layer image is directly used as the first layer image of the current scanning layer; or, If the current scanning plane is the first type of scanning plane in the next scanning plane segment, the original K-space data is supplemented and collected according to the sampling rate of the first type of scanning plane, and a first-level image of the first type of scanning plane is generated based on the supplemented K-space data.

6. The method according to any one of claims 1-5, characterized in that, The step of determining the adjacent scanning layers corresponding to the current scanning layer further includes: Obtain the prediction mode of the current scanning layer, wherein the prediction mode includes one-way inter-frame prediction and two-way inter-frame prediction; Based on the prediction pattern, the adjacent scanning layers are determined by the scanning layers on which the prediction of the current scanning layer depends.

7. The method according to claim 6, characterized in that, The step of determining the adjacent scanning layers based on the prediction mode for the current scanning layer includes: If the prediction mode is a one-way inter-frame prediction mode, the forward prediction scanning layer on which the current scanning layer image is predicted is used as the adjacent scanning layer. If the prediction mode is the bidirectional inter-frame prediction mode, the forward prediction scanning layer and the backward prediction scanning layer on which the current scanning layer image is predicted are determined as the adjacent scanning layers.

8. The method according to claim 1, characterized in that, Also includes: If the current scanning plane is a set of planes including multiple sub-scanning planes, starting from the first sub-scanning plane in the set, MR inter-frame prediction mode is performed on each sub-scanning plane in the set to generate a second-level image corresponding to each sub-scanning plane in the set, until the imaging of each sub-scanning plane in the set is completed, and then the next scanning plane or the next scanning plane segment is entered.

9. The method according to any one of claims 1-5, characterized in that, The process of generating a second-level image of the current scanned layer includes: The feature extraction layer extracts features from the input layer image and outputs a feature image. The input layer image is the first layer image and the second layer image of the adjacent scanning layer when performing MR inter-frame prediction mode, and the first layer image when performing MR intra-frame imaging. The data processing layer generates predicted K-space data from the feature image, and based on the predicted K-space data, generates a predicted enhancement layer image for the current scanning layer. The predicted enhancement layer image is input into the image enhancement layer, residual information is output, and a second layer image of the current scan layer is generated based on the residual information and the predicted enhancement layer image.

10. The method according to claim 9, characterized in that, The step of generating a predicted augmented layer image for the current scanned layer based on the predicted K-space data includes: Replace the predicted data at the sampling frequency in the predicted K-space data with the original K-space data to generate the target K-space data; A predicted augmented layer image of the current scanned layer is generated based on the target K-space data.

11. The method according to claim 10, characterized in that, Intra-MR imaging is performed by an intra-MR imaging model, and inter-MR prediction mode is performed by an inter-MR prediction mode model. Both the inter-MR prediction mode model and the intra-MR imaging model include multiple image processing units connected sequentially. Each image processing unit includes a feature extraction layer, a data processing layer, and an image enhancement layer. The feature extraction layer, the data processing layer, and the image enhancement layer within each image processing unit are connected sequentially, and the output of the image enhancement layer is the input of the feature extraction layer in the next image processing unit.

12. A scanning device for magnetic resonance imaging (MRI), characterized in that, include: The generation module is used to acquire the original K-space data of the current scanning plane in each scanning plane segment, and generate the first plane image of the current scanning plane based on the original K-space data; The determination module is used to determine whether the current scanning layer is a first type of scanning layer, wherein the first type of scanning layer uses an intra-frame imaging mode for imaging; The acquisition module is configured to, in response to the current scanning layer not being a first type of scanning layer, determine the adjacent scanning layer corresponding to the current scanning layer and acquire a second layer image of the adjacent scanning layer, wherein the resolution of the second layer image is greater than the resolution of the first layer image. An imaging module is used to predict and image the current scanning layer based on the first layer image and the second layer image of the adjacent scanning layer, using an MR inter-frame prediction mode, to generate the second layer image of the current scanning layer. The imaging module is further configured to, if the current scanning plane is the first type of scanning plane, perform intra-MR imaging on the current scanning plane based on the first plane image to generate a second plane image of the current scanning plane.

13. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-11.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-11.

15. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-11.

Images