Magnetic resonance imaging method and device and computer equipment

Abstract

The invention relates to a magnetic resonance imaging method and device and computer equipment. The method comprises the following steps: based on a partition undersampling mode, obtaining a plurality of radially directed blades rotating around a K space center; each blade is provided with a plurality of partitions distributed in the radial direction; the subarea undersampling mode comprises that the undersampling degree corresponding to each subarea is gradually increased along the direction far away from the center of the K space; and performing image reconstruction according to the plurality of leaves to obtain a magnetic resonance image. According to the method, the sampling of each blade is ensured to be different by utilizing a partition undersampling mode, so that the whole K space is covered to the maximum extent, the higher recovery acceleration capability is obtained based on the repeated acquisition characteristic, and the high-acceleration imaging from sequence acquisition to reconstruction is realized.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a magnetic resonance imaging method, apparatus and computer equipment. Background Art

[0002] At present, MRI is slow, and patients cannot remain still for a long time, which can easily cause motion artifacts in the image. Solutions to this problem include using propeller imaging, but traditional image processing methods used to reduce artifacts will increase scanning time. Summary of the invention

[0003] Based on this, it is necessary to provide a magnetic resonance imaging method, apparatus and computer equipment that can reduce scanning time in response to the above technical problems.

[0004] In a first aspect, the present application provides a magnetic resonance imaging method, the method comprising:

[0005] Based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are obtained; the blade has a plurality of radially distributed partitions; the partition undersampling method includes gradually increasing the degree of undersampling corresponding to each partition in a direction away from the center of the K space;

[0006] Image reconstruction is performed based on multiple leaves to obtain a magnetic resonance image.

[0007] In one embodiment, the partition includes a first region located at a central position of the K space and a second region located at a peripheral position of the K space;

[0008] Based on the partition undersampling method, multiple radially pointing blades rotating around the center of K space are obtained, including:

[0009] Undersampling the first region and the second region according to a set of radial K-space sampling trajectories to obtain a magnetic resonance signal; wherein the acceleration rate of the second region is greater than the acceleration rate of the first region;

[0010] The magnetic resonance signal is filled into the K space to obtain the blade.

[0011] In one embodiment, the method further comprises:

[0012] dividing the second area into a third area adjacent to the first area and a fourth area adjacent to the third area;

[0013] The data filling densities of the first region, the third region, and the fourth region are successively sparse.

[0014] In one embodiment, the method further comprises:

[0015] dividing the second area into a third area adjacent to the first area, a fourth area adjacent to the third area, and a fifth area adjacent to the fourth area;

[0016] The acceleration rates of the first region, the third region, the fourth region and the fifth region increase in multiples.

[0017] In one embodiment, based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are acquired, including:

[0018] The sampling points in the same partition are collected in a random under-sampling mode.

[0019] In one embodiment, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline.

[0020] In one embodiment, performing image reconstruction based on multiple leaves to obtain a magnetic resonance image includes:

[0021] Acquiring first coil sensitivity information of a plurality of blade overlapping regions;

[0022] Fill each leaf according to the sensitivity information of the first coil to obtain each filled leaf;

[0023] Acquire the second coil sensitivity information of each filled leaf, and obtain an initialization result based on the second coil sensitivity information;

[0024] Iterative reconstruction is performed according to the initialization result to obtain a magnetic resonance image.

[0025] In one of the embodiments, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline;

[0026] Iterative reconstruction is performed according to the initialization results to obtain a magnetic resonance image, including:

[0027] Using the half-Fourier reconstruction method, the second set of data corresponding to the leaf with low sampling density and located on the side of the center line of the K space is fitted according to the first set of data to complete the single leaf initialization;

[0028] According to the leaf data after the single leaf initialization is completed, the sensitivity of the receiving coil is obtained;

[0029] According to the sensitivity of the receiving coil, the blade data is restored secondary by using a parallel reconstruction method and a neural network combination method to obtain synthetic data.

[0030] In a second aspect, the present application also provides a magnetic resonance imaging device, comprising:

[0031] An undersampling module is used to obtain a plurality of radially directed blades rotating around the center of the K space based on a partitioned undersampling method; the blade has a plurality of radially distributed partitions; the partitioned undersampling method includes gradually increasing the undersampling degree corresponding to each partition along a direction away from the center of the K space;

[0032] The image reconstruction module is used to perform image reconstruction according to the multiple leaves to obtain a magnetic resonance image.

[0033] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0034] The magnetic resonance imaging method, apparatus and computer equipment described above obtain multiple radially directed blades rotating around the center of the K space based on a partitioned undersampling method; wherein the blade has multiple radially distributed partitions, and the partitioned undersampling method includes gradually increasing the degree of undersampling corresponding to each partition along the direction away from the center of the K space; and then reconstructing the image based on the multiple blades to obtain a magnetic resonance image. The present application uses a partitioned undersampling method to ensure that the sampling of each blade is different, so as to cover the entire K space to the maximum extent, obtain a higher recovery acceleration capability based on the repeated acquisition characteristics, and realize high-acceleration imaging from sequence acquisition to reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A diagram showing an application environment of a magnetic resonance imaging method in an embodiment;

[0036] Figure 2 is a schematic flow chart of a magnetic resonance imaging method in one embodiment;

[0037] Figure 3 A schematic diagram of a propeller in one embodiment;

[0038] Figure 4 It is a schematic diagram of a process of acquiring leaves by using a zone undersampling method in one embodiment;

[0039] Figure 5 A schematic diagram of a K-space sampling trajectory in an embodiment;

[0040] Figure 6 A schematic diagram of a K-space sampling trajectory in another embodiment;

[0041] Figure 7 is a schematic diagram of a process of image reconstruction in one embodiment;

[0042] Figure 8 A schematic diagram of a specific process of image reconstruction in one embodiment;

[0043] Fig.9A A trajectory set diagram obtained after the first blade trajectory rotates in one embodiment;

[0044] Fig. 9B A trajectory set diagram obtained after the second blade trajectory rotates in one embodiment;

[0045] Fig.10 is a structural block diagram of a magnetic resonance imaging device in one embodiment;

[0046] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0049] It is to be understood that terms such as “first”, “second”, etc. in the present application are only used to distinguish similar objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features.

[0050] It can be understood that the “connection” in the following embodiments should be understood as “electrical connection”, “communication connection”, etc. if the connected circuits, modules, units, etc. have electrical signals or data transmission between each other.

[0051] It will be understood that “at least one” means one or more, and “plurality” means two or more.

[0052] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0053] Motion artifacts are prone to occur during scanning in magnetic resonance imaging. Traditional schemes such as propeller imaging can reduce artifacts caused by rigid body motion, but at the same time the scanning time will be increased. Currently, among the acceleration schemes commonly used in propeller technology, parallel imaging uses irrelevant information between multiple channels for acceleration, which is limited by the design and data of the coil and has limited acceleration capabilities. Some Fourier techniques use the conjugate symmetry characteristics of K space for acceleration, but in actual use they are limited by phase changes, which affects the phase correction results in propeller technology and limits their use. Regarding deep learning, the current combination of propeller technology and deep learning mostly uses deep learning networks to obtain single-piece images and motion information. Similar information between blades is not fully utilized, and most of the K space sampling trajectories are random sampling, and the initialization results are prone to artifacts.

[0054] The magnetic resonance imaging method provided in this application can be applied to Figure 1 In the application environment shown. The application environment is a magnetic resonance system, which includes a bed 110 and an MR scanner 120. The MR scanner 120 can perform sampling on the detection object (a part of the detection object 010), obtain the magnetic resonance signal of the detection object, and perform Fourier transformation on the magnetic resonance signal to obtain a magnetic resonance image of the detection object.

[0055] The MR scanner 120 includes a magnet, a radio frequency transmitting coil, a gradient coil, and a radio frequency receiving coil. The bed 110 is used to carry the detection object 010, the radio frequency transmitting coil is used to transmit radio frequency pulses to the detection object, the gradient coil is used to generate a gradient field, and the gradient field can be along the phase encoding direction, the layer selection direction or the frequency encoding direction, etc.; the radio frequency receiving coil is used to receive magnetic resonance signals. In one embodiment, the magnet of the MR scanner 120 can be a permanent magnet or a superconducting magnet, and according to different functions, the radio frequency coils constituting the radio frequency unit can be divided into a body coil and a local coil. In one embodiment, the types of the radio frequency transmitting coil and the radio frequency receiving coil can be a birdcage coil, a solenoid coil, a saddle coil, a Helmholtz coil, an array coil, a loop coil, etc.

[0056] The magnetic resonance system may also include an output device, such as a display, which can display the magnetic resonance image of the region of interest. Further, the output device can also display the height, weight, age, imaging part of the subject, and the working status of the MR scanner 120. The type of the output device can be a cathode ray tube (CRT) output device, a liquid crystal output device (LCD), an organic light emitting output device (OLED), a plasma output device, etc., or a combination of several of them.

[0057] In one embodiment, Figure 2 As shown, a magnetic resonance imaging method is provided, which is applied to Figure 1 The nuclear magnetic resonance system in FIG. 1 is taken as an example to illustrate the method, which includes the following steps:

[0058] Step 202, based on the partitioned undersampling method, a plurality of radially directed blades rotating around the center of the K space are obtained; the blades have a plurality of radially distributed partitions; the partitioned undersampling method includes gradually increasing the degree of undersampling corresponding to each partition along the direction away from the center of the K space.

[0059] Among them, the blade is explained as a technical term in the embodiment of the present application as follows. It should be noted that even if a name is adopted in the following description, this name is only for the unified description of the whole text and is only used as a reference. In fact, there are other possible names, which will not be repeated here: the magnetic resonance K-space blade acquired according to the PROPELLER sequence. The basic sequence of the propeller (PROPELLER, periodically rotated overlapping parallel lines with enhanced reconstruction) technology is a fast spin echo sequence (FSE, Fast Spin Echo) or a fast inversion recovery sequence (FIR, Fast Inversion Recovery), which has an echo chain. An echo chain is collected in a repetition period (TR, Repetition Time). Each echo in the echo chain needs to be frequency encoded and phase encoded, and filled in parallel in the K space at a certain angle. This set of filling information can be called a blade or data strip of PROPELLER.

[0060] like Figure 3 As shown, the complete propeller consists of a group of blades at a certain angle to each other. Each blade contains multiple parallel data lines, and each data line extends along the phase encoding direction. The blades can be set to rotate and collect at an angle of 180° / N, where N is an integer and the number of blades. For example, a golden angle of 137.51° or a half golden angle of 68.75° can be set to rotate and collect to form multiple blades, and multiple blades are arranged in an interlaced manner.

[0061] Exemplarily, the present application proposes to adopt a partitioned undersampling method to obtain multiple radially directed blades rotating around the center of the K space, so that the blades have multiple radially distributed partitions. In the embodiment of the present application, the partitioned undersampling method can refer to an independently designed K space sampling scheme for each blade of the propeller technology, thereby ensuring that the sampling of each blade is different to maximize the coverage of the entire K space, wherein the partitioning can ensure the stability of the prior knowledge required for subsequent reconstruction, while providing greater acceleration possibilities, thereby utilizing the unique repeated acquisition characteristics of the propeller technology to obtain a higher recovery acceleration capability.

[0062] Specifically, the partitioned undersampling method refers to the design of K-space sampling for partitioned individual leaves, and then the leaves have multiple radially distributed partitions, which can refer to dividing the independent leaves into N regions (N is a positive integer, and the value of N can be a range). For these N regions, the partitioned undersampling method includes gradually increasing the degree of undersampling / sparseness corresponding to each partition along the direction away from the center of the K-space, that is, from the radial direction pointing from the center of the K-space to the edge of the K-space, the sampling of the K-space gradually becomes sparse, optimizing the sampling density, thereby avoiding acceleration artifacts caused by sampling. Compared with the Cartesian sampling trajectory, the present application can achieve imaging with a higher acceleration multiple.

[0063] Taking the example of dividing an independent leaf into N regions, the first region is the K space region, which uses a low undersampling degree; the second region is the two side regions near the K space, and its undersampling degree can be higher than the first region; the third region is the outer side of the second region away from the center of the K space, and so on. This application optimizes the sampling density to obtain a higher recovery acceleration capability while avoiding acceleration artifacts caused by sampling.

[0064] Step 204: reconstruct an image based on the multiple leaves to obtain a magnetic resonance image.

[0065] Specifically, when multiple blades are acquired by partition undersampling, image reconstruction can be performed based on the multiple blades to obtain a magnetic resonance image. For example, all blades can be simultaneously transmitted to a deep learning network for reconstruction, and then the repeated acquisition characteristics of propeller technology can be used to obtain a higher recovery acceleration capability. The embodiment of the present application has a higher acceleration capability than the current propeller acceleration solution.

[0066] It should be noted that the partitioned undersampling method and reconstruction scheme proposed in the present application are also applicable to non-propeller technology scenarios, for example, they can be applied to other conventional sequence acquisitions, and the embodiments of the present application are not limited to this.

[0067] The above-mentioned magnetic resonance imaging method utilizes partitioned undersampling to ensure that the sampling of each leaf is different, so as to cover the entire K space to the maximum extent, obtain higher recovery acceleration capability based on the repeated acquisition characteristics, and realize high-acceleration imaging from sequence acquisition to reconstruction.

[0068] In one embodiment, the partition includes a first region located at the center of the K space and a second region located at the peripheral position of the K space; Figure 4 As shown, based on the partition undersampling method, multiple radially directed blades rotating around the center of the K space are obtained, including:

[0069] Step 302, undersampling the first region and the second region according to a set of radial K-space sampling trajectories to obtain a magnetic resonance signal; wherein the acceleration rate of the second region is greater than the acceleration rate of the first region;

[0070] Specifically, in the embodiment of the present application, the blade has multiple radially distributed partitions, wherein the partition may include a first region located at the center of the K space, and a second region located at the peripheral position of the K space; the nuclear magnetic resonance system can undersample the first region and the second region with different undersampling ratios according to a set of radial K space sampling trajectories, thereby obtaining a magnetic resonance signal.

[0071] By dividing the independent leaves into the first area and the second area, the stability of the prior knowledge required for subsequent reconstruction can be guaranteed. By designing the sampling in which the acceleration factor of the second area is greater than the acceleration factor of the first area, it can be ensured that the sampling of each leaf is different, thereby maximizing the coverage of the entire K space and realizing a high acceleration scheme to reduce the scanning time.

[0072] Step 304: fill the magnetic resonance signal into the K space to obtain the blade.

[0073] Specifically, when the magnetic resonance signal is obtained based on the partition undersampling method, the magnetic resonance signal can be filled into the K space to obtain the blade. Among them, a set of radial K space sampling trajectories can correspond to a blade, and another set of radial K space sampling trajectories obtained by rotating a set of radial K space sampling trajectories by a certain angle can be used to obtain another blade, and this process is repeated until the K space is filled, that is, multiple blades form a complete propeller.

[0074] In one embodiment, the method may further include:

[0075] dividing the second area into a third area adjacent to the first area and a fourth area adjacent to the third area;

[0076] The data filling densities of the first region, the third region, and the fourth region are successively sparse.

[0077] Specifically, regarding the design of the K-space sampling trajectory, the embodiment of the present application proposes that a three-partition method (i.e., N=3) can be adopted. Through the partition sampling method, the independent leaves are divided into three regions. The first region is the K-space region, and the third region is the two side regions near the K-space. The fourth region is the outside of the third region away from the center of the K-space, and so on.

[0078] Among them, the data filling density of the first area, the third area and the fourth area is sparse in turn, so that the embodiment of the present application can optimize the sampling density while increasing the acceleration possibility through the three-partition method, avoiding acceleration artifacts caused by sampling and improving image quality.

[0079] In one embodiment, the method may further include:

[0080] dividing the second area into a third area adjacent to the first area, a fourth area adjacent to the third area, and a fifth area adjacent to the fourth area;

[0081] The acceleration rates of the first region, the third region, the fourth region and the fifth region increase in multiples.

[0082] Specifically, regarding the design of the K-space sampling trajectory, the embodiment of the present application proposes that a 4-partition method (i.e., N=4) can be adopted, where the first area is the K-space area, and the third area is the two side areas near the K-space, the fourth area is the outside of the third area away from the center of the K-space, and the fifth area is the outside of the fourth area away from the center of the K-space, and so on.

[0083] Furthermore, the acceleration rates of the first region, the third region, the fourth region and the fifth region increase in multiples; illustratively, the first region is undersampled by 2 times, the third region is undersampled by 3 times, and the density of the fifth region is twice that of the fourth region. Figure 5 The figure shows a schematic diagram of the filling of the collected magnetic resonance signal in the K space. Each row is a sampling data line of an independent leaf. The closer the area is to the center of the K space, the denser the sampling density (corresponding to the bright spot in the figure) and the lower the degree of undersampling; the farther away from the center of the K space, the sparser the sampling density and the higher the degree of undersampling. While the embodiment of the present application increases the possibility of greater acceleration through the 4-partition method, it optimizes the sampling density by increasing the undersampling ratio in multiples, which can effectively avoid acceleration artifacts caused by sampling and ensure image quality.

[0084] In one embodiment, based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are acquired, including:

[0085] The sampling points of the first area are collected in an irregular under-sampling mode or a regular under-sampling mode.

[0086] Specifically, for the first area located at the center of the K space, a low-magnification undersampling method is adopted, and regular or irregular sampling can be performed; among them, for the first area, the irregular sampling mode is adopted to reduce the generation of artifacts and improve the spatial resolution, while the regular sampling mode is adopted to collect the central part of the K space, thereby ensuring image contrast and maintaining image resolution.

[0087] The above-mentioned magnetic resonance imaging method ensures the stability of the prior knowledge required for subsequent reconstruction through partitioning, while providing greater acceleration possibilities. It optimizes the sampling density to avoid acceleration artifacts caused by sampling, and then utilizes the unique repeated acquisition characteristics of propeller technology to obtain higher recovery acceleration capabilities and ensure image quality.

[0088] In one embodiment, based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are acquired, including:

[0089] The sampling points in the same partition are collected in a random under-sampling mode.

[0090] Specifically, in the embodiment of the present application, the random sampling mode means that regular undersampling is not performed, and some distribution functions can be used randomly. Random processing can be performed between sampling points in the same partition to further optimize the sampling density and avoid acceleration artifacts caused by regular sampling.

[0091] In one embodiment, based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are acquired, including:

[0092] The overall sampling density of the sampling points located on one side of the center line of the K space is greater than the overall sampling density of the sampling points located on the other side of the center line of the K space.

[0093] Specifically, in the embodiment of the present application, the overall sampling density of the K space may not be symmetrical along the center of the K space, wherein the overall sampling density on one side of the K space may be greater than the overall sampling density on the other side. Furthermore, in the embodiment of the present application, the acceleration capability of the approximate partial Fourier method can be obtained in the scenario where the linear order of the K space needs to be maintained, but the influence of the partial Fourier algorithm on the phase is avoided, and the subsequent reconstruction algorithm may cause the failure of the phase correction algorithm required in the propeller technology.

[0094] In one embodiment, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline.

[0095] Specifically, in the embodiment of the present application, the K-space trajectory can be further optimized so that the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline, thereby reducing image blur caused by insufficient high-frequency information in the K-space. Among them, the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline can refer to the left side of the K-space, and the K-space sampling trajectory on the other side of the K-space centerline can refer to the right side of the K-space.

[0096] For example, Figure 6The K-space sampling trajectory design shown in the figure, each row is a schematic diagram of the sampling points of an independent leaf. The embodiment of the present application proposes that the sampling density on the left side of the K-space can be reduced, and the sampling of high-frequency components can be greatly reduced, but still partially retained. Then, the highly accelerated partial data can be iteratively calculated through partial Fourier technology, parallel imaging and deep learning joint methods, thereby reducing the image blur caused by insufficient high-frequency information in the K-space when only partial Fourier technology is used in actual use.

[0097] In one embodiment, Figure 7 As shown, image reconstruction is performed based on multiple leaves to obtain a magnetic resonance image, including:

[0098] Step 402, obtaining first coil sensitivity information of a plurality of blade overlapping regions;

[0099] Specifically, when multiple blades are acquired, the nuclear magnetic resonance system can estimate coil sensitivity information 1 (first coil sensitivity information), wherein the coil sensitivity information 1 can be preliminarily calculated using data repeatedly acquired at the center of the K space.

[0100] Exemplarily, obtaining the first coil sensitivity information of the overlapping region of multiple blades may refer to the MRI system estimating the coil sensitivity information 1 using the filling signal of the central part of the K space obtained by overlapping all blades. Optionally, the coil sensitivity information 1 may also be obtained by Figure 3 The data filled in the overlapping area of ​​two adjacent leaves, three leaves or multiple leaves as shown is obtained by calculation, and the embodiments of the present application are not limited to this.

[0101] Step 404, filling each leaf according to the sensitivity information of the first coil to obtain each filled leaf;

[0102] Specifically, in the case of coil sensitivity information 1 (first coil sensitivity information) obtained by calculation, the nuclear magnetic resonance system can preliminarily fill each blade with the coil sensitivity information 1. Optionally, the coil sensitivity information is used for data supplementation, and more information can be supplemented between multiple blades, thereby coping with the situation in which the data used to calculate the coil sensitivity information in the propeller technology is very limited.

[0103] Exemplarily, the nuclear magnetic resonance system can fill the part of each blade with acceleration less than 4 times by using the obtained coil sensitivity information 1, thereby enhancing the signal and optimizing the image quality.

[0104] Step 406, obtaining the second coil sensitivity information of each filled leaf, and obtaining an initialization result based on the second coil sensitivity information;

[0105] Specifically, when obtaining each filled blade, the nuclear magnetic resonance system can use the filled K space to recalculate the coil sensitivity information 2 (second coil sensitivity information) of each blade; illustratively, the nuclear magnetic resonance system can recalculate the coil sensitivity information 2 of the filled independent blade, and then use it for subsequent fine calculations.

[0106] Step 408: Perform iterative reconstruction according to the initialization result to obtain a magnetic resonance image.

[0107] Specifically, after obtaining the initialization result, the MRI system can iteratively reconstruct according to the initialization result to obtain a magnetic resonance image. For example, after obtaining the detailed calculation results, the MRI system can simultaneously reconstruct all blades by passing them into the deep learning network, and use the repeated acquisition characteristics unique to propeller technology to obtain a higher recovery acceleration capability, thereby having a higher acceleration capability than the current acceleration solution of propeller technology.

[0108] Optionally, after the nuclear magnetic resonance system recalculates the coil sensitivity information 2 of each leaf using the filled K space, the information can be used to calculate the initialization result of each leaf, and the result can be sent to the network for iterative reconstruction after phase correction and motion correction, as shown in the following formula:

[0109]

[0110] Among them, argmin means the minimum value operation; represents the operation of finding the two-norm; x is the target solution data, corresponding to the target image after the iteration; y represents the actual collected data, which is undersampled along the phase encoding direction; is the constraint term introduced by AI, x AI It is the data obtained by neural network prediction. The constraint items make the target contour corresponding to the prediction of the neural network consistent with the contour corresponding to the actually collected data. E represents the data recovery mode corresponding to the actually collected data. The data recovery mode can be, for example, partial Fourier reconstruction, reconstruction based on compressed sensing, and parallel reconstruction of K space.

[0111] like Figure 8 A schematic diagram of acquiring a magnetic resonance image based on multiple leaves is shown, which may include:

[0112] 1) Data collection process: In this embodiment, eight leaves of k-space are collected according to the PROPELLER scheme. The k-space data lines contained in the eight leaves approximately cover the central area of ​​the k-space. The leaves of the k-space rotate around the center of the k-space, wherein each leaf of the k-space collects at least a part of the central part of the k-space, and the sampling density of each leaf along the radial direction of the k-space is gradually sparse, that is, each leaf of the k-space is under-sampled;

[0113] 2) A data recovery process including parallel imaging reconstruction, in this embodiment: first, a first group of K-space data lines included in two adjacent leaves or multiple adjacent leaves can be obtained; then, according to the central area of ​​the K-space corresponding to the first group of K-space data lines, the sensitivity 1 of the receiving coil is calculated; then, according to the sensitivity 1 of the receiving coil, a parallel reconstruction method such as SMASH, AUTO-SMASH, VD-AUTO-SMASH or GRAPPA is used to fit the intra-leaf data of at least one of the two adjacent leaves or multiple adjacent leaves and the uncollected data between the leaves to complete the single-leaf initialization; the single-leaf initialization can be performed for the above-mentioned eight leaves respectively; further, based on the sensitivity 1 of the receiving coil, According to the initialized blade data, the sensitivity 2 of the receiving coil can be calculated. In this embodiment, the eight initialized blades can be divided into multiple groups adjacent to each other. According to the central area of ​​the k-space corresponding to each group of k-space data lines, the intermediate sensitivities of the multiple receiving coils are calculated, and the intermediate sensitivities of the multiple receiving coils are averaged or correlated to obtain the sensitivity 2 of the receiving coil; finally, the sensitivity 2 of the receiving coil and the data after the single blade is initialized are input into the network (neural network) to obtain synthetic data, and the synthetic data includes the data collected during the data collection process and the restored data of the unsampled points; or, according to the sensitivity 2 of the receiving coil, a half-Fourier (Partial Fourier, PF) reconstruction method is used to fit the part of the blade with low sampling density to obtain an intermediate fitting data set, and the intermediate fitting data set is input into the network to obtain synthetic data.

[0114] The above steps may be repeated multiple times until the number of cycles meets a set threshold, or the image quality corresponding to the synthesized data meets clinical requirements.

[0115] Optionally, the training process of the neural network may include: obtaining a series of sample under-sampled K-space data sets and corresponding sample coil sensitivity distributions as input samples; outputting the sample full-sampled K-space data sets corresponding to the series of sample under-sampled K-space data sets as the gold standard, and continuously adjusting the loss function of the neural network to obtain a trained deep learning model.

[0116] 3) Phase correction and motion correction are performed on the synthesized data in the K-space domain. The motion correction may include: obtaining motion parameters corresponding to each blade; correcting each blade image based on the motion parameters; merging each corrected blade image to obtain an intermediate magnetic resonance image, and transforming the intermediate magnetic resonance image to the K-space domain to obtain corrected data. The motion parameters can be obtained by: correcting each blade according to the PROPELLER algorithm, calculating the data changes caused by rigid body motion between the blades, and calculating the motion calibration coefficient of each blade as the motion parameter.

[0117] 4) Grid the corrected data. Grid the data of multiple blades together to the target blade coordinate system grid, and then uniformly extract the data in the direction of the missing data of the target blade for calculation.

[0118] In one embodiment, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline;

[0119] Iterative reconstruction is performed according to the initialization results to obtain a magnetic resonance image, including:

[0120] Acquire a first set of data corresponding to a leaf with a high sampling density and located on one side of the center line of the K space;

[0121] Using the Partial Fourier (PF) reconstruction method, the second set of data corresponding to the leaf with low sampling density and located on the side of the K-space centerline is fitted according to the first set of data to complete the single leaf initialization;

[0122] According to the initialized blade data, the sensitivity 2 of the receiving coil can be calculated. In this embodiment, the eight initialized blades can be divided into multiple groups adjacent to each other. According to the central area of ​​the k-space corresponding to each group of k-space data lines, the intermediate sensitivities of the multiple receiving coils are calculated. The intermediate sensitivities of the multiple receiving coils are averaged or correlated to obtain the sensitivity 2 of the receiving coil.

[0123] According to the sensitivity 2 of the receiving coil, the blade data is restored twice by using a parallel reconstruction method and a neural network combination method to obtain synthetic data.

[0124] The above-mentioned magnetic resonance imaging method obtains multiple radially directed blades rotating around the center of K space based on a partitioned undersampling method; wherein the blade has multiple radially distributed partitions, and the partitioned undersampling method includes gradually increasing the degree of undersampling corresponding to each partition along the direction away from the center of K space; and then reconstructing the image based on the multiple blades to obtain a magnetic resonance image. The present application uses a partitioned undersampling method to ensure that the sampling of each blade is different, so as to cover the entire K space to the maximum extent, obtain a higher recovery acceleration capability based on the repeated acquisition characteristics, and realize high-acceleration imaging from sequence acquisition to reconstruction.

[0125] In high-field imaging, the B1 field is relatively non-uniform due to physical reasons such as the skin effect. To overcome this, pTX emission is usually required, which will result in an exponential increase in imaging time.

[0126] In one embodiment, multiple excitation modes are used, and different encoding trajectories are used for leaves at the same encoding position in different excitation modes, including:

[0127] In the first excitation mode, a first blade trajectory for K-space acquisition is set; wherein the first blade trajectory is set to be along a direction away from the center of the K-space, and a plurality of partitions are set, and the undersampling degree corresponding to each of the partitions gradually increases;

[0128] The sequence is executed according to the trajectory of the first blade, and the first blade is rotated and collected at a set angle of 180° / N to obtain a first set of data sets, where N is an integer; Fig.9A is a trajectory of the first blade and a trajectory set diagram obtained after its rotation;

[0129] In the second excitation mode, a second blade trajectory for K-space acquisition is set; wherein the second blade trajectory is set to be in a direction away from the center of the K-space, a plurality of partitions are set, and the undersampling degree corresponding to each of the partitions gradually increases, the first blade trajectory and the second blade trajectory correspond to the same phase encoding position, but the trajectory is set differently (indicated by different lines in the figure);

[0130] The sequence is executed according to the second blade trajectory, and the second blade is rotated at a set angle of 180° / N to obtain a second set of data sets, where N is an integer; Fig. 9B is a trajectory of the second blade and a trajectory set diagram obtained after the second blade rotates;

[0131] Reconstructing the first set of data and the second set of data to obtain a first magnetic resonance image and a second magnetic resonance image respectively;

[0132] The first magnetic resonance image and the second magnetic resonance image are fused to obtain a target magnetic resonance image.

[0133] In the embodiment of the present application, by setting different tracks of the same encoding position under different excitation modes, the amount of non-redundant information can be increased, and more image information can be provided when multi-emission mode data is merged.

[0134] In another embodiment, a multiple excitation mode is used, each excitation corresponds to a leaf in the K space, and the sampling trajectories of two adjacent leaves are different:

[0135] Setting a first blade trajectory for K-space acquisition; wherein the first blade trajectory is set to be along a direction away from the center of the K-space, and multiple sub-areas are set, and the undersampling degree corresponding to each of the sub-areas gradually increases;

[0136] Execute the sequence according to the first blade trajectory to obtain a first blade data set;

[0137] A second blade trajectory for K-space acquisition is set, and the second blade has a set angle with the first blade; wherein the second blade trajectory is set in a direction away from the center of the K-space, a plurality of partitions are set, and the undersampling degree corresponding to each of the partitions gradually increases, and the first blade trajectory and the second blade trajectory correspond to the same phase encoding position, but the setting trajectories are different;

[0138] Execute the sequence according to the second blade trajectory to obtain a second blade data set;

[0139] Image reconstruction is performed according to the first leaf data set and the second leaf data set to obtain a magnetic resonance image.

[0140] In the embodiment of the present application, by setting different trajectories for different blades, the amount of non-redundant information can be increased, and more image information can be provided when the multi-emission mode data is merged. It should be understood that although the various steps in the flowcharts involved in the embodiments described above are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise clearly stated in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowcharts involved in the embodiments described above may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed in turn or alternating with other steps or at least part of the steps or stages in other steps.

[0141] Based on the same inventive concept, the embodiment of the present application also provides a magnetic resonance imaging device for implementing the magnetic resonance imaging method involved above. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above method, so the specific limitations in one or more magnetic resonance imaging device embodiments provided below can refer to the limitations on the magnetic resonance imaging method above, and will not be repeated here.

[0142] In one embodiment, Fig.10 As shown, a magnetic resonance imaging device is provided, comprising:

[0143] The undersampling module 1001 is used to obtain a plurality of radially directed blades rotating around the center of the K space based on a partitioned undersampling method; the blade has a plurality of radially distributed partitions; the partitioned undersampling method includes gradually increasing the undersampling degree corresponding to each partition along a direction away from the center of the K space;

[0144] The image reconstruction module 1002 is used to perform image reconstruction according to multiple leaves to obtain a magnetic resonance image.

[0145] In one embodiment, the partition includes a first area located at the center of the K space and a second area located at the peripheral position of the K space; the undersampling module 1001 includes

[0146] A sampling module, used for undersampling the first region and the second region according to a set of radial K-space sampling trajectories to obtain a magnetic resonance signal; wherein the acceleration rate of the second region is greater than the acceleration rate of the first region;

[0147] The signal filling module is used to fill the magnetic resonance signal into the K space to obtain the blade.

[0148] In one embodiment, the device further comprises:

[0149] The area division module is used to divide the second area into a third area adjacent to the first area and a fourth area adjacent to the third area; the data filling density of the first area, the third area and the fourth area is sparse in sequence.

[0150] In one embodiment, the device further comprises:

[0151] The area division module is used to divide the second area into a third area adjacent to the first area, a fourth area adjacent to the third area, and a fifth area adjacent to the fourth area; the acceleration rates of the first area, the third area, the fourth area and the fifth area increase in multiples.

[0152] In one embodiment, the undersampling module 1001 is used to collect sampling points in the same partition in a random undersampling mode.

[0153] In one embodiment, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline.

[0154] In one embodiment, the image reconstruction module 1002 includes:

[0155] A first sensitivity information acquisition module, used to acquire first coil sensitivity information of a plurality of blade overlap areas;

[0156] An information filling module, used to fill each leaf according to the sensitivity information of the first coil to obtain each filled leaf;

[0157] A second sensitivity acquisition module, used to acquire the second coil sensitivity information of each filled leaf, and obtain an initialization result based on the second coil sensitivity information;

[0158] The iterative reconstruction module is used to perform iterative reconstruction according to the initialization result to obtain a magnetic resonance image.

[0159] In one of the embodiments, the sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline;

[0160] The iterative reconstruction module is used to adopt a half-Fourier reconstruction method to fit a second set of data located on one side of the K-space centerline corresponding to the blade with low sampling density according to the first set of data to complete the single-leaf initialization; obtain the sensitivity of the receiving coil according to the blade data after the single-leaf initialization; and according to the sensitivity of the receiving coil, use a parallel reconstruction method and a neural network combination method to perform secondary restoration on the blade data to obtain synthetic data.

[0161] Each module in the above-mentioned magnetic resonance imaging device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0162] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.11As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface, the display unit and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a magnetic resonance imaging method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0163] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of the above-mentioned magnetic resonance imaging method when executing the computer program.

[0165] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the magnetic resonance imaging method described above are implemented.

[0166] In one embodiment, a computer program product is provided, comprising a computer program, which implements the steps of the magnetic resonance imaging method described above when executed by a processor.

[0167] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0168] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0169] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A magnetic resonance imaging method, characterized in that: The method comprises: Based on the partition undersampling method, a plurality of radially directed blades rotating around the center of the K space are obtained; the blades have a plurality of radially distributed partitions; the partition undersampling method includes gradually increasing the degree of undersampling corresponding to each partition in a direction away from the center of the K space; Image reconstruction is performed according to the multiple leaves to obtain a magnetic resonance image.

2. The method according to claim 1, characterized in that The partition includes a first area located at a central position of the K space and a second area located at a peripheral position of the K space; The method of acquiring multiple radially directed blades rotating around the center of the K space based on the partition undersampling method includes: Undersampling the first region and the second region according to a set of radial K-space sampling trajectories to obtain a magnetic resonance signal; wherein the acceleration rate of the second region is greater than the acceleration rate of the first region; The magnetic resonance signal is filled into the K space to obtain the blade.

3. The method according to claim 2, characterized in that The method further comprises: dividing the second area into a third area adjacent to the first area and a fourth area adjacent to the third area; The data filling densities of the first region, the third region and the fourth region are successively sparse.

4. The method according to claim 2, characterized in that: The method further comprises: dividing the second area into a third area adjacent to the first area, a fourth area adjacent to the third area, and a fifth area adjacent to the fourth area; The acceleration rates of the first region, the third region, the fourth region and the fifth region increase in multiples.

5. The method according to claim 1, characterized in that The method of acquiring a plurality of radially directed blades rotating around the center of the K space based on the partition undersampling method includes: The sampling points in the same partition are collected in a random under-sampling mode.

6. The method according to claim 1, characterized in that The sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline.

7. The method according to claim 1, characterized in that The step of reconstructing an image based on the plurality of blades to obtain a magnetic resonance image comprises: Acquiring first coil sensitivity information of the overlapping area of ​​the plurality of blades; Filling each of the leaves according to the sensitivity information of the first coil to obtain each filled leaf; Acquiring the second coil sensitivity information of each of the filled leaves, and obtaining an initialization result based on the second coil sensitivity information; Iterative reconstruction is performed according to the initialization result to obtain the magnetic resonance image.

8. The method according to claim 7, characterized in that The sampling density of the K-space sampling trajectory corresponding to the blade on one side of the K-space centerline is greater than the sampling density of the K-space sampling trajectory on the other side of the K-space centerline; The iterative reconstruction according to the initialization result to obtain the magnetic resonance image includes: Acquire a first set of data corresponding to the leaf with high sampling density and located on one side of the center line of the K space; Using a half-Fourier reconstruction method, a second set of data corresponding to the leaf with low sampling density and located on one side of the center line of the K space is fitted according to the first set of data to complete the single leaf initialization; Obtaining the sensitivity of the receiving coil according to the leaf data after the single leaf initialization is completed; According to the sensitivity of the receiving coil, the blade data is restored twice by adopting a parallel reconstruction method and a neural network combination method to obtain synthetic data.

9. A magnetic resonance imaging device, characterized in that: The device comprises: An undersampling module is used to obtain a plurality of radially directed blades rotating around the center of the K space based on a partitioned undersampling method; the blade has a plurality of radially distributed partitions; the partitioned undersampling method includes gradually increasing the undersampling degree corresponding to each partition in a direction away from the center of the K space; An image reconstruction module is used to perform image reconstruction according to the multiple leaves to obtain a magnetic resonance image.

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

Images