Magnetic resonance imaging method, device, equipment and medium

Abstract

Embodiments of the present invention disclose a magnetic resonance imaging method, apparatus, device, and medium. The method includes: acquiring unfilled K-space data to be processed in a high-frequency region; performing super-resolution reconstruction on the K-space data to obtain K-space data of a super-resolution image; zero-filling the K-space data to be processed to generate zero-filled reconstructed K-space data consistent with the K-space range of the K-space data of the super-resolution image; calculating, based on an original image corresponding to the K-space data to be processed, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image; generating merged K-space data based on the zero-filled reconstructed K-space data processed based on the first weight matrix and the K-space data of the super-resolution reconstructed image processed based on the second weight matrix; and reconstructing the merged K-space data to generate a target magnetic resonance image. The weight matrix is ​​automatically calculated to obtain a high-resolution magnetic resonance image.

Description

Technical Field

[0001] The embodiments of the present invention relate to image processing technology, and in particular to a magnetic resonance imaging method, apparatus, device and medium. Background Art

[0002] A typical magnetic resonance imaging system consists of the following components: a magnet, gradient coils, a radiofrequency (RF) transmit coil, a RF receive coil, and a signal processing and image reconstruction unit. The spin of hydrogen nuclei in the human body can be described as a small magnetic needle. Within the strong magnetic field provided by the magnet, the hydrogen nuclei shift from a chaotic thermal equilibrium state to one that is partially aligned with and partially opposed to the main magnetic field. The difference between the two forms the net magnetization vector. The hydrogen nuclei precess around the main magnetic field, with the precession frequency proportional to the magnetic field strength. The gradient unit generates a magnetic field whose strength varies with spatial position, which is used for spatial encoding of the signal. The RF transmit coil flips the hydrogen nuclei from the main magnetic field to a transverse plane, causing them to precess around the main magnetic field. This induces a current signal in the RF receive coil. Signal processing and image reconstruction units produce an image of the tissue being imaged.

[0003] In the above process, the processed magnetic resonance signals need to be filled into K space before image reconstruction. The filling of K space is closely related to the speed of magnetic resonance imaging. The filling speed of K space is actually consistent with the acquisition speed of magnetic resonance signals. Once the K space is filled with data, the acquisition of magnetic resonance signals is also complete. The number of sampling points in the frequency encoding direction of K space is consistent with the actual number of pixels in the frequency encoding direction of the magnetic resonance image. The number of points in the phase encoding direction of K space (the number of phase encoding lines) is also consistent with the number of pixels in the phase encoding direction of the magnetic resonance image. Therefore, the larger the dot matrix of K space, the smaller the image pixels and the higher the spatial resolution, but the more acquisition time is required.

[0004] The general process of MRI involves pre-scanning, positioning setup, sequence setup, gradient setup, RF signal transmission, coil reception of feedback signals, and post-processing of the signals back to the workstation to generate the final scan signal. Each step requires processing time, and because MRI sequences are numerous, multiple scans are required for each site. For example, a cranial MRI scan requires T1, T2, mass, hydrops, and arterial and venous imaging sequences. Each scan takes longer than a single CT scan. Consequently, obtaining high-resolution MRI images requires not only a long scan waiting period but also the patient's relative immobility during the scan. However, due to individual patient variability, such as physical condition and tolerance, obtaining high-resolution MRI images can be challenging. Therefore, improvements to existing MRI methods are necessary to reduce scan times while still achieving the desired high-resolution images. Summary of the Invention

[0005] Embodiments of the present invention provide a magnetic resonance imaging method, apparatus, device, and medium to implement K-space data merging based on an automatically calculated weight matrix, thereby obtaining a magnetic resonance image with more details and higher resolution based on the merged K-space data.

[0006] In a first aspect, an embodiment of the present invention provides a magnetic resonance imaging method, the method comprising:

[0007] Acquiring K-space data to be processed, wherein a high-frequency region of the K-space data to be processed is not filled;

[0008] Performing super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image;

[0009] Filling an unfilled area of ​​the to-be-processed K-space data with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has a consistent K-space range with the K-space data of the super-resolution image;

[0010] Calculate, based on the original image, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image;

[0011] Merging the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data;

[0012] The combined K-space data is reconstructed to generate a target magnetic resonance image.

[0013] In a second aspect, an embodiment of the present invention further provides a magnetic resonance imaging apparatus, the apparatus comprising:

[0014] A K-space data acquisition module is used to acquire K-space data to be processed, wherein the high-frequency area of ​​the K-space data to be processed is not filled;

[0015] A super-resolution reconstruction module is used to perform super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image;

[0016] a zero-filling reconstructed K-space data generating module, configured to fill in zeros in an unfilled area of ​​the to-be-processed K-space data to generate zero-filling reconstructed K-space data, wherein the zero-filling reconstructed K-space data has the same K-space range as the K-space data of the super-resolution image;

[0017] A weight matrix calculation module, configured to calculate, based on the original image, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image;

[0018] A K-space data merging module is configured to merge the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data;

[0019] The image generation module is used to reconstruct the combined K-space data and generate a target magnetic resonance image.

[0020] In a third aspect, an embodiment of the present invention further provides an imaging device, wherein the imaging device includes:

[0021] one or more processors;

[0022] a storage device for storing one or more programs;

[0023] When the one or more programs are executed by the one or more processors, the one or more processors implement the magnetic resonance imaging method provided by any embodiment of the present invention.

[0024] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the magnetic resonance imaging method provided by any embodiment of the present invention.

[0025] The embodiment of the present invention obtains K-space data to be processed, wherein the high-frequency region of the K-space data to be processed is unfilled; performs super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image; the super-resolution image improves image sharpness; fills the unfilled region of the K-space data to be processed with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has the same K-space range as the K-space data of the super-resolution image; obtains K-space data of a larger matrix based on the K-space data to be processed; calculates a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image; merges the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; and reconstructs the merged K-space data to generate a target magnetic resonance image. This solves the problem of being unable to automatically calculate the weight matrix required for K when merging spatial data, and achieves automatic calculation of the weight matrix, thereby obtaining a magnetic resonance image with greater detail and higher resolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of a magnetic resonance imaging method in embodiment 1 of the present invention;

[0027] Figure 2 is a schematic diagram of a voxel in the first embodiment of the present invention;

[0028] Figure 3 Schematic diagram of K space in two-dimensional imaging in embodiment 1 of the present invention;

[0029] Figure 4 is a schematic diagram of the in-plane resolution of a voxel in the first embodiment of the present invention;

[0030] Figure 5 Schematic diagram of generating super-resolution reconstructed K-space data and zero-filled reconstructed K-space data in the first embodiment of the present invention;

[0031] Figure 6 2 is a schematic diagram of merging the zero-filled reconstructed K-space data and the K-space data of the super-resolution reconstructed image in the first embodiment of the present invention;

[0032] Figure 7 is a structural diagram of a magnetic resonance imaging device in embodiment 2 of the present invention;

[0033] Figure 8 It is a structural schematic diagram of an imaging device in embodiment 3 of the present invention. DETAILED DESCRIPTION

[0034] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0035] Example 1

[0036] Figure 1 This is a flow chart of a magnetic resonance imaging method provided in the first embodiment of the present invention. This embodiment is applicable to magnetic resonance imaging. The method can be performed by a magnetic resonance imaging device and specifically includes the following steps:

[0037] S110 , obtaining K-space data to be processed, wherein the high-frequency region of the K-space data to be processed is not filled.

[0038] The K-space data to be processed is the K-space data that has not been reconstructed with super-resolution. In addition, the high-frequency area of ​​the K-space data to be processed is not filled.

[0039] In one embodiment, the K-space data to be processed is obtained in the following manner: the matrix size corresponding to the K-space is 256 (frequency encoding) × 256 (phase encoding), and the K-space data to be processed is under-collected along the phase encoding direction, thereby forming a matrix of 256 (frequency encoding) × 128 (phase encoding), that is, the data lines in the phase encoding direction are reduced, and the already collected data lines preferentially fill the central / center area of ​​the K-space.

[0040] In one embodiment, the K-space data to be processed is obtained by setting a preset weight curve and determining the number of acquisitions required for each phase encoding layout in K-space based on the preset weight curve. In this embodiment, the central region of K-space is assigned the highest weight, i.e., the number of repeated acquisitions is the greatest. The weights gradually decrease from the central region to the peripheral regions, and the number of acquisitions in the peripheral regions of K-space is zero. In this manner, the K-space data to be processed can be obtained.

[0041] The filling trajectory of the data lines of the to-be-processed K-space data in the K-space can be, for example, line-by-line filling, circuitous filling, spiral filling, or radial filling along the phase encoding direction. In this embodiment, there is no specific limitation on the filling trajectory of the data lines in the K-space.

[0042] S120 , performing super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image.

[0043] The K-space data to be processed is super-resolution reconstructed to obtain K-space data of a higher frequency super-resolution image. Optionally, an artificial neural network can be trained by collecting low-resolution sample data as input and high-resolution sample data as expected output. The artificial neural network is composed of a large number of nodes (or neurons) connected to each other. Each node represents a specific output function, called an activation function. The connection between each two nodes represents a weighted value for the signal passing through the connection, called a weight. The artificial neural network includes a data input layer, an intermediate hidden layer, and a data output layer. The artificial neural network model used in the embodiment of the present invention can be a convolutional neural network (CNN), a generative adversarial network (GAN) or other forms of neural network models.

[0044] The generated K-space data is transformed into the image domain. The sample data generated by the low-resolution K-space is used as input, and the sample data generated by the high-resolution K-space is used as the expected output to train the artificial neural network, thereby obtaining a model that can be used to achieve super-resolution reconstruction. Optionally, the cost function can be selected as the mean square error function, as follows:

[0045] C(w,b)=∑||y(a)-z|| 2

[0046] Where a is the input of the super-resolution reconstruction model, y(a) is the output of the super-resolution reconstruction model, and z is the expected output of the super-resolution reconstruction model. 2 Represents a two-norm operation, used to calculate the loss of the cost function. w and b are two variables, and changes in the values ​​of the two variables will affect the loss of the cost function. In this embodiment, w and b are the weight and bias of the network, respectively.

[0047] The k-space data to be processed is input into a trained super-resolution reconstruction model to obtain k-space data for a higher super-resolution image. Optionally, sensitivity encoding (SENSE), simultaneous acquisition of spatial harmony (SMASH), or generalized auto-calibrating partially parallel acquisition (GRAPPA) can be used to obtain k-space data corresponding to a super-resolution image with a higher resolution than the k-space data to be processed.

[0048] The resolution of a magnetic resonance image refers to the size of the spatial voxel represented by each pixel in the reconstructed image. Figure 2 In one embodiment, the K-space data to be processed corresponds to a two-dimensional image, and its resolution includes: the in-plane resolution Δx, Δy of the voxel, and the slice thickness. Figure 3 The figure shows the K space diagram in two-dimensional imaging. The K space data to be processed are filled in the frequency encoding direction Kx and the phase encoding direction Ky. Among them, the resolution in the X direction Δx = 1 / (2·k x，max ), the resolution in the Y direction Δy=1 / (2·k y，max ), k x，max Indicates the highest frequency of the sampling point along the frequency encoding direction, k y，max Indicates the highest frequency of sampling points along the phase encoding direction. In other words, the higher the frequency of K-space acquisition points (acquisition frequency), the smaller the voxels represented by the reconstructed image pixels, and the higher the spatial resolution.

[0049] In one embodiment, the K-space data to be processed corresponds to a three-dimensional image, and the resolution includes three values: the in-plane resolution of the voxel Δx, Δy, and Δz. Figure 4 Figure 2 shows a schematic diagram of the in-plane resolution of a voxel. According to the Nyquist sampling theorem, the resolution Δx in the frequency encoding direction is determined by the highest frequency of the sampling points in K-space along this frequency encoding direction; the resolution Δy in the frequency encoding direction is determined by the highest frequency of the sampling points in K-space along the first phase encoding direction; and the resolution Δz in the frequency encoding direction is determined by the highest frequency of the sampling points in K-space along the second phase encoding direction. The first phase encoding direction is perpendicular to the second phase encoding direction; specifically, the first phase encoding direction is the intra-plane phase encoding direction, while the second phase encoding direction is the inter-plane phase encoding direction.

[0050] S130 , filling an unfilled area of ​​the K-space data to be processed with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has a consistent K-space range with the K-space data of the super-resolution image.

[0051] In order to obtain a larger matrix of K-space data based on the K-space data to be processed, a zero-filling operation is performed on the unfilled area of ​​the K-space data to be processed, and the missing K-space data is filled with zeros to generate zero-filled reconstructed K-space data. Figure 5 The figure shows the super-resolution reconstructed K-space data and the zero-filled reconstructed K-space data obtained based on the K-space data to be processed, i.e., the original K-space data. However, the zero-filled reconstructed K-space data must have the same size as the matrix formed by the K-space data of the super-resolution image, otherwise the zero-filled reconstructed K-space data cannot be merged with the K-space data of the super-resolution image. In an embodiment of the present invention, the unfilled area of ​​the K-space data to be processed is filled with zeros. Specifically, before the K-space data is reconstructed into an image, the center of the K-space is filled with the actually acquired data, and the surrounding K-space is partially filled with zeros, thereby obtaining a 512 or 1024 matrix. The above method can be referred to as an "intra-layer zero-filled interpolation method" or an "intra-layer zero-filled difference method." Correspondingly, the "zero-filled reconstructed K-space data" can also be referred to as "zero-filled extended K-space data."

[0052] S140 , calculating, based on the original image, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image.

[0053] When merging the zero-filled reconstructed K-space data with the K-space data of the super-resolution image, it is necessary to calculate a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image. The zero-filled reconstructed K-space data and the K-space data of the super-resolution image are merged using the first and second weight matrices to generate merged K-space data.

[0054] Optionally, calculating a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image corresponding to the K-space data to be processed includes: calculating the signal-to-noise ratio of the original image; and calculating the first weight matrix and the second weight matrix based on the signal-to-noise ratio. In 2D multi-layer acquisition or 3D acquisition, the first weight matrix and the second weight matrix are calculated separately for different layers, i.e., different weight matrices may be applied to images in different layers.

[0055] Optionally, calculating the signal-to-noise ratio of the original image includes: calculating the noise level of the original image; transforming the K-space data of the original image into image space to obtain image space data; dividing each pixel value in the image space data by the noise level to obtain the signal-to-noise ratio distribution / signal-to-noise ratio distribution graph of the original image; obtaining the signal-to-noise ratio exceeding a preset threshold value based on the signal-to-noise ratio distribution; and calculating the average signal-to-noise ratio corresponding to pixels whose noise levels exceed the preset threshold value to obtain the signal-to-noise ratio of the original image. Using the average signal-to-noise ratio corresponding to pixels whose noise levels exceed the preset threshold value as the signal-to-noise ratio of the original image can make the obtained signal-to-noise ratio of the original image more accurate. Exemplarily, calculating the average signal-to-noise ratio value of pixels in the signal-to-noise ratio distribution that are significantly higher than the noise level, for example, more than 6 times, is the signal-to-noise ratio value to be obtained.

[0056] Optionally, calculating the noise level of the original image includes: obtaining a preset number of data point information in edge rows and edge columns of the K-space data of the original image, and calculating the noise level based on the data point information. Using M complex data points at the edge of the K-space data to be processed, i.e., corresponding to the highest frequency component, where M is a natural number greater than 1, the noise level is estimated according to the following formula:

[0057]

[0058] Among them, x j It is the real part or imaginary part of M complex data points, a total of 2M data. is the mean of the real and imaginary parts of all data points, j is the number of the complex data point, and 1≤j≤2M.

[0059] The signal-to-noise ratio of the original image is obtained by calculating the noise level of the original image, and then the first weight matrix and the second weight matrix are obtained according to the signal-to-noise ratio.

[0060] Optionally, calculating the noise level of the original image further includes: acquiring a preset amount of noisy K-space data; and calculating the noise level based on the noisy K-space data. Using the same sequence used to acquire the original image, with the RF pulse turned off, several sets of pure noise data with no signal are acquired, and the noise level is estimated according to the above formula to further calculate the signal-to-noise ratio of the original image. In this embodiment of the present application, the signal-to-noise ratio of the original image can also be referred to as the "quality factor" of the original image.

[0061] Optionally, calculating the signal-to-noise ratio of the original image further includes: inputting the K-space data of the original image into a trained signal-to-noise ratio calculation model to obtain the signal-to-noise ratio of the original image. Obtaining the image signal-to-noise ratio using an artificial intelligence method: obtaining historical image data or historical K-space data labeled with image signal-to-noise ratio values ​​for training a neural network for signal-to-noise ratio calculation. Directly inputting the image data or K-space data for which the signal-to-noise ratio is to be calculated into a trained neural network model for signal-to-noise ratio calculation, outputting the required signal-to-noise ratio, and then calculating a first weight matrix and a second weight matrix based on the output signal-to-noise ratio.

[0062] Optionally, the signal-to-noise ratio of the original image can also be obtained by calculating the degree of artifacts in the original image. Image artifacts herein include but are not limited to motion artifacts, radio frequency interference artifacts, artifacts caused by reconstruction distortion, and the like.

[0063] Optionally, the first weight matrix and the second weight matrix are calculated according to the signal-to-noise ratio, including: inputting the signal-to-noise ratio into a preset monotonically increasing function to obtain a first weight matrix; inputting the signal-to-noise ratio into a preset monotonically decreasing function to obtain a second weight matrix. Exemplarily, the obtained signal-to-noise ratio value of the original image is input into a preset monotonically increasing function such as W1=Q^2 to obtain a first weight matrix. Wherein W1 is the first weight matrix, and Q is the signal-to-noise ratio of the original image. The obtained signal-to-noise ratio value of the original image is input into a preset monotonically decreasing function such as W2=1 / Q^2 to obtain a second weight matrix. Wherein W2 is the second weight matrix, and Q is the signal-to-noise ratio of the original image. The above method can automatically calculate the first weight matrix corresponding to the zero-filled reconstructed K-space data and the second weight matrix corresponding to the K-space data of the super-resolution image without the need for setting through experience, thereby improving the merging quality and efficiency of the K-space data and making the obtained merged K-space data more in line with actual needs.

[0064] S150 , merging the zero-filled reconstructed K-space data processed based on the first weight matrix and the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data.

[0065] like Figure 6Figure 2 shows a schematic diagram of merging zero-filled reconstructed K-space data with K-space data from a super-resolution reconstructed image. The zero-filled reconstructed K-space data is multiplied by a first weight matrix, and the K-space data from the super-resolution image is multiplied by a second weight matrix. The zero-filled reconstructed K-space data multiplied by the first weight matrix and the K-space data from the super-resolution image multiplied by the second weight matrix are then merged to produce merged K-space data.

[0066] S160: Reconstruct the merged K-space data to generate a target magnetic resonance image.

[0067] The merged K-space data is reconstructed to obtain a target magnetic resonance image. Compared with the original image, the obtained target magnetic resonance image has more details and higher resolution. More lesion or human tissue information can be obtained through the image, which is more clinically valuable.

[0068] The technical solution of this embodiment comprises: obtaining K-space data to be processed, wherein the high-frequency region of the K-space data to be processed is unfilled; super-resolution reconstructing the K-space data to be processed to obtain K-space data of a super-resolution image; the super-resolution image improves image sharpness; filling the unfilled regions of the K-space data to be processed with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has the same K-space range as the K-space data of the super-resolution image; obtaining K-space data of a larger matrix based on the K-space data to be processed; calculating a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image; merging the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; and reconstructing the merged K-space data to generate a target magnetic resonance image. This solves the problem of being unable to automatically calculate the weight matrix required for merging K-space data, thereby achieving automatic calculation of the weight matrix, thereby obtaining a magnetic resonance image with greater detail and higher resolution.

[0069] Example 2

[0070] Figure 7 This is a structural diagram of a magnetic resonance imaging device provided in Example 3 of the present invention, which includes: a K-space data acquisition module 310, a super-resolution reconstruction module 320, a zero-filling reconstruction K-space data generation module 330, a weight matrix calculation module 340, a K-space data merging module 350 and an image generation module 360.

[0071] Among them, the K-space data acquisition module 310 is used to obtain K-space data to be processed, and the high-frequency area of ​​the K-space data to be processed is not filled; the super-resolution reconstruction module 320 is used to perform super-resolution reconstruction on the K-space data to be processed to obtain K-space data of the super-resolution image; the zero-filled reconstructed K-space data generation module 330 is used to fill the unfilled area of ​​the K-space data to be processed with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data is consistent with the K-space range of the K-space data of the super-resolution image; the weight matrix calculation module 340 is used to calculate a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image corresponding to the K-space data to be processed; the K-space data merging module 350 is used to merge the zero-filled reconstructed K-space data processed based on the first weight matrix and the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; the image generation module 360 ​​is used to reconstruct the merged K-space data to generate a target magnetic resonance image.

[0072] In the technical solution of the above embodiment, the weight matrix calculation module 340 includes:

[0073] a signal-to-noise ratio calculation unit, configured to calculate the signal-to-noise ratio of the original image;

[0074] A weight matrix calculation unit is used to calculate the first weight matrix and the second weight matrix according to the signal-to-noise ratio.

[0075] In the technical solution of the above embodiment, the signal-to-noise ratio calculation unit includes:

[0076] a noise level calculation subunit, configured to calculate the noise level of the original image;

[0077] A K-space data conversion subunit, configured to convert the K-space data of the original image into image space to obtain image space data;

[0078] a signal-to-noise ratio distribution generating subunit, configured to divide each pixel value in the image spatial data by the noise level to obtain a signal-to-noise ratio distribution of the original image;

[0079] a signal-to-noise ratio acquisition subunit, configured to acquire a signal-to-noise ratio exceeding a preset threshold according to the signal-to-noise ratio distribution;

[0080] The signal-to-noise ratio calculation subunit is configured to calculate an average of the signal-to-noise ratios corresponding to pixels whose noise levels exceed a preset threshold to obtain the signal-to-noise ratio of the original image.

[0081] In the technical solution of the above embodiment, the noise level calculation subunit includes:

[0082] The data point information acquisition subunit is used to acquire a preset number of data point information in edge rows and edge columns in the K-space data of the original image, and calculate the noise level according to the data point information.

[0083] In the technical solution of the above embodiment, the noise level calculation subunit further includes:

[0084] The noise K-space data acquisition subunit is used to acquire a preset amount of noise K-space data and calculate the noise level according to the noise K-space data.

[0085] In the technical solution of the above embodiment, the signal-to-noise ratio calculation unit further includes:

[0086] The original image input subunit is used to input the K-space data of the original image into the trained signal-to-noise ratio calculation model to obtain the signal-to-noise ratio of the original image.

[0087] In the technical solution of the above embodiment, the weight matrix calculation unit includes:

[0088] A first weight matrix generating subunit, configured to input the signal-to-noise ratio into a preset monotonically increasing function to obtain a first weight matrix;

[0089] The second weight matrix generating subunit is configured to input the signal-to-noise ratio into a preset monotonically decreasing function to obtain a second weight matrix.

[0090] The technical solution of this embodiment comprises: obtaining K-space data to be processed, wherein the high-frequency region of the K-space data to be processed is unfilled; super-resolution reconstructing the K-space data to be processed to obtain K-space data of a super-resolution image; the super-resolution image improves image sharpness; filling the unfilled regions of the K-space data to be processed with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has the same K-space range as the K-space data of the super-resolution image; obtaining K-space data of a larger matrix based on the K-space data to be processed; calculating a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image; merging the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; and reconstructing the merged K-space data to generate a target magnetic resonance image. This solves the problem of being unable to automatically calculate the weight matrix required for merging K-space data, thereby achieving automatic calculation of the weight matrix, thereby obtaining a magnetic resonance image with greater detail and higher resolution.

[0091] The magnetic resonance imaging apparatus provided by the embodiment of the present invention can execute the magnetic resonance imaging method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0092] Example 3

[0093] Figure 8 A structural diagram of an imaging device provided in the third embodiment of the present invention is shown in FIG. Figure 8 As shown, the imaging device includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of the processor 410 in the imaging device can be one or more. Figure 8 In the figure, a processor 410 is used as an example; the processor 410, the memory 420, the input device 430 and the output device 440 in the imaging device can be connected via a bus or other means. Figure 8 The bus connection is taken as an example.

[0094] The memory 420, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the magnetic resonance imaging method in the embodiments of the present invention (e.g., the K-space data acquisition module 310, the super-resolution reconstruction module 320, the zero-filled reconstructed K-space data generation module 330, the weight matrix calculation module 340, the K-space data merging module 350, and the image generation module 360 ​​in the magnetic resonance imaging device). The processor 410 executes the software programs, instructions, and modules stored in the memory 420 to execute various functional applications and data processing of the imaging device, thereby implementing the magnetic resonance imaging method described above.

[0095] Memory 420 may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal. Furthermore, memory 420 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state memory device. In some instances, memory 420 may further include memory remotely located relative to processor 410, and such remote memory may be connected to the imaging device via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0096] The input device 430 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the imaging device. The output device 440 may include a display device such as a display screen.

[0097] Example 4

[0098] A fourth embodiment of the present invention further provides a storage medium comprising computer-executable instructions, wherein the computer-executable instructions, when executed by a computer processor, are used to perform a magnetic resonance imaging method, the method comprising:

[0099] Acquiring K-space data to be processed, wherein a high-frequency region of the K-space data to be processed is not filled;

[0100] Performing super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image;

[0101] Filling an unfilled area of ​​the to-be-processed K-space data with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has a consistent K-space range with the K-space data of the super-resolution image;

[0102] Calculate, based on the original image, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image;

[0103] Merging the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data;

[0104] The combined K-space data is reconstructed to generate a target magnetic resonance image.

[0105] Of course, the computer-executable instructions of a storage medium provided by an embodiment of the present invention are not limited to the operations of the method described above, but can also execute related operations in the magnetic resonance imaging method provided by any embodiment of the present invention.

[0106] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0107] It is worth noting that in the above-mentioned embodiment of the magnetic resonance imaging apparatus, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0108] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A magnetic resonance imaging method, characterized in that: include: Acquiring K-space data to be processed, wherein a high-frequency region of the K-space data to be processed is not filled; Performing super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image; Filling an unfilled area of ​​the to-be-processed K-space data with zeros to generate zero-filled reconstructed K-space data, wherein the zero-filled reconstructed K-space data has a consistent K-space range with the K-space data of the super-resolution image; Calculating a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on the original image corresponding to the K-space data to be processed; Merging the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; reconstructing the combined K-space data to generate a target magnetic resonance image; The calculating, based on the original image corresponding to the K-space data to be processed, a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image includes: Calculating the signal-to-noise ratio of the original image; Calculating the first weight matrix and the second weight matrix according to the signal-to-noise ratio; The calculating the first weight matrix and the second weight matrix according to the signal-to-noise ratio includes: Inputting the signal-to-noise ratio into a preset monotonically increasing function to obtain a first weight matrix; The signal-to-noise ratio is input into a preset monotonically decreasing function to obtain a second weight matrix.

2. The method according to claim 1, characterized in that Calculating the signal-to-noise ratio of the original image includes: Calculating the noise level of the original image; Converting the K-space data of the original image into image space to obtain image space data; Dividing each pixel value in the image spatial data by the noise level to obtain a signal-to-noise ratio distribution of the original image; Acquire a signal-to-noise ratio exceeding a preset threshold according to the signal-to-noise ratio distribution; The signal-to-noise ratio of the original image is obtained by calculating an average value of the signal-to-noise ratios corresponding to pixels whose noise levels exceed a preset threshold.

3. The method according to claim 2, characterized in that The calculating the noise level of the original image includes: A preset number of data point information in edge rows and edge columns in the K-space data of the original image is obtained, and a noise level is calculated based on the data point information.

4. The method according to claim 2, characterized in that The calculating the noise level of the original image further includes: A preset amount of noise K-space data is acquired, and a noise level is calculated according to the noise K-space data.

5. The method according to claim 1, wherein The calculating the signal-to-noise ratio of the original image further includes: The K-space data of the original image is input into a trained signal-to-noise ratio calculation model to obtain the signal-to-noise ratio of the original image.

6. A magnetic resonance imaging apparatus, characterized in that: include: A K-space data acquisition module is used to acquire K-space data to be processed, wherein the high-frequency area of ​​the K-space data to be processed is not filled; A super-resolution reconstruction module is used to perform super-resolution reconstruction on the K-space data to be processed to obtain K-space data of a super-resolution image; a zero-filling reconstructed K-space data generating module, configured to fill in zeros in an unfilled area of ​​the to-be-processed K-space data to generate zero-filling reconstructed K-space data, wherein the zero-filling reconstructed K-space data has the same K-space range as the K-space data of the super-resolution image; a weight matrix calculation module, configured to calculate a first weight matrix corresponding to the zero-filled reconstructed K-space data and a second weight matrix corresponding to the K-space data of the super-resolution image based on an original image corresponding to the K-space data to be processed; A K-space data merging module is configured to merge the zero-filled reconstructed K-space data processed based on the first weight matrix with the K-space data of the super-resolution reconstructed image processed based on the second weight matrix to generate merged K-space data; An image generation module, configured to reconstruct the combined K-space data and generate a target magnetic resonance image; The weight matrix calculation module includes: a signal-to-noise ratio calculation unit, configured to calculate the signal-to-noise ratio of the original image; a weight matrix calculation unit, configured to calculate the first weight matrix and the second weight matrix according to the signal-to-noise ratio; The weight matrix calculation unit includes: A first weight matrix generating subunit, configured to input the signal-to-noise ratio into a preset monotonically increasing function to obtain a first weight matrix; The second weight matrix generating subunit is configured to input the signal-to-noise ratio into a preset monotonically decreasing function to obtain a second weight matrix.

7. An imaging device, characterized in that The imaging device comprises: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the magnetic resonance imaging method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the magnetic resonance imaging method according to any one of claims 1 to 5 is implemented.

Images