Three-dimensional fast magnetic resonance elastography acquisition and reconstruction method and system

Abstract

This invention provides a three-dimensional fast magnetic resonance elastography acquisition and reconstruction method and system, comprising: Step S1: generating vibration using an external vibrator and transmitting it into human tissue to generate fluctuations within the tissue; Step S2: having the magnetic resonance scanner receive a synchronization signal and applying an image acquisition sequence to record the fluctuation information of the human tissue; Step S3: acquiring the raw data and using a fast reconstruction algorithm to obtain an accurate fluctuation image of the human tissue. This invention uses a multi-block alternating acquisition method, which can maintain the signal-to-noise ratio of the image even with short transmission cycles (TRs), thereby increasing the number of samples within the same vibration cycle and improving acquisition efficiency. This invention uses three-dimensional excitation and acquisition, increasing the intensity of the acquired signal itself, thus overcoming the shortcomings of previous methods.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional dynamic magnetic resonance imaging technology, specifically, to a method and system for three-dimensional fast magnetic resonance elastography acquisition and reconstruction; more specifically, to a method and system for three-dimensional fast magnetic resonance elastography acquisition and reconstruction based on multi-layer block alternating acquisition and Hadamard encoding. Background Technology

[0002] Existing technologies have the following drawbacks: In conventional magnetic resonance imaging, the repetition time (TR) cannot be set too short due to the limitation of the longitudinal magnetization vector's recovery time; otherwise, the image signal-to-noise ratio will be reduced. This limits the number of time points that can be recorded within the same vibration cycle when using multi-phase displacement encoded stimulated echoes to record wave images, thus affecting acquisition efficiency.

[0003] Some existing patents rely on displacement-encoded stimulated echo recording of wave images, which, while shortening scan time, leads to a decrease in the acquired signal strength, thus degrading the quality of the target wave image. Some patents use planar echo imaging (EPI) or spiral acquisition methods, which significantly accelerate acquisition time but introduce image artifacts, affecting imaging accuracy. Existing patents require six encoding steps to acquire three-directional wave images, reducing imaging speed. Most existing patents use two-dimensional Cartesian sampling, which is easily affected by motion and cannot achieve high undersampling ratios. Existing patents are limited by the recovery time of the longitudinal magnetization vector, requiring a long repetition time, which impacts scan speed.

[0004] Existing technologies include: Strasser J, Haindl MT, Stollberger R, et al. Magneticresonance elastography of the human brain using a multiphase DENSEacquisition[J]. Magnetic Resonance in Medicine, 2019, 81(6): 3578-3587. and Garteiser P, Sahebjavaher RS, Ter Beek LC, et al. Rapid acquisition of multifrequency, multislice and multidirectional MR elastography data with afractionally encoded gradient echo sequence[J].NMR in Biomedicine, 2013, 26(10):1326-1335.

[0005] Patent document CN106264530B discloses a Cartesian k-space acquisition method and system for three-dimensional dynamic magnetic resonance imaging. The method includes: establishing a k-space model in a three-dimensional Cartesian coordinate system; determining the acquisition trajectory of echo signals in the model; wherein the acquisition trajectory of the echo signals is such that all echo signals are acquired parallel along one coordinate direction, and the coordinates of each echo signal in the plane formed by the other two coordinate directions are obtained by the two-dimensional golden ratio; determining the scanning time series and the encoding gradient of the magnetic field to be applied by the magnetic resonance imaging system based on the acquisition trajectory; setting the magnetic resonance imaging system according to the time series and encoding gradient, and acquiring k-space data conforming to the acquisition trajectory. However, this invention cannot achieve a high undersampling factor. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for three-dimensional rapid magnetic resonance elastography acquisition and reconstruction.

[0007] A three-dimensional fast magnetic resonance elastography acquisition and reconstruction method according to the present invention includes:

[0008] Step S1: Use an external vibrator to generate vibrations and transmit them into the human tissue, creating fluctuations within the tissue.

[0009] Step S2: Instruct the magnetic resonance scanner to receive a synchronization signal and apply an image acquisition sequence to record the fluctuation information of human tissue;

[0010] Step S3: Obtain the raw data and use a fast reconstruction algorithm to obtain an accurate wave image in human tissue.

[0011] Preferably, in step S2:

[0012] Step S2.1: At the beginning of the sequence, apply an electric field with intensities K in three directions. x ,K y ,K z The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton;

[0013] Step S2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional block, and the position of the layer direction is encoded using a spatial phase-coded gradient. Two adjacent regions to be imaged are alternately excited, using pulses with intensities K in three directions. x ,K y ,K z The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; step S2.2 is a motion decoding module;

[0014] Step S2.3: If full sampling is performed, change the spatial phase encoding gradient in the motion decoding module and repeat it N times, where N is the number of target layers required, until the same golden angle k spatial radial trajectory line in all target layers is collected. In the N repeated collections, the spatial phase encoding gradient is set to first the center and then the two sides.

[0015] If undersampling is performed, then... Second collection.

[0016] Preferably, step S2.4: repeat steps S2.1 to S2.3, changing the collected golden angle k-space radial trajectory line in each repetition;

[0017] If full sampling is performed, steps S2.1 to S2.3 are repeated M times, where M is related to the resolution; if the target resolution is m×m, then

[0018] If undersampling is performed, repeat steps S2.1 to S2.3. In this case, R represents the undersampling factor;

[0019] Step S2.5: Use Hadamard encoding to record the fluctuation information in different directions; repeat step S2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are:

[0020] D1 = [+1, -1, -1]

[0021] D2 = [+1,+1,+1]

[0022] D3 = [-1, -1, +1]

[0023] D4 = [-1, +1, -1]

[0024] The order is not limited, where +1 represents positive and -1 represents negative.

[0025] Preferably, in step S3:

[0026] Step S3.1: Arrange the original data of the two adjacent regions that need to be imaged according to their true k-space positions, and fill the missing positions due to undersampling with 0;

[0027] Step S3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image:

[0028]

[0029] Where x is the target fluctuation image to be recovered, y is the acquired raw data, S is the coil sensitivity map, and F... u It is the Fourier transform and undersampling operator. It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform, and λ1 and λ2 are the weights that control the two regularization constraints, respectively.

[0030] Preferably, step S3.3: Equation (1) is iteratively optimized using the alternating direction multiplier method until the objective function converges, resulting in the reconstructed complex image x of the two blocks. slab1 With x slab2 Calculate the phase φ of the complex image respectively, and obtain φ slab1 With φ slab2 ;

[0031] Step S3.4: Decode the phase image φ, which is caused by the superposition of multi-directional information due to Hadamard encoding, using equation (2):

[0032]

[0033]

[0034] Where, * represents dot product, indicating multiplication at corresponding positions; φ0 is the background phase outside the target wave pattern; E is the Hadamard coding matrix, which is related to the applied gradient direction and is constructed based on the actual direction; φ1, φ2, φ3, and φ4 are the phase patterns obtained by gradient coding using four directions: D1, D2, D3, and D4, respectively; the resulting u = [u x ,u y ,u z ] T For the target fluctuation image.

[0035] A three-dimensional fast magnetic resonance elastography acquisition and reconstruction system according to the present invention includes:

[0036] Module M1: Uses an external vibrator to generate vibrations and transmits them into human tissue, creating ripples within the tissue.

[0037] Module M2: Enables the magnetic resonance scanner to receive a synchronization signal and applies an image acquisition sequence to record the wave information of human tissue;

[0038] Module M3: Acquires raw data and uses a fast reconstruction algorithm to obtain accurate wave images of human tissue.

[0039] Preferably, in module M2:

[0040] Module M2.1: At the beginning of the sequence, apply an electric field with intensities K in three directions. x ,K y ,K z The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton;

[0041] Module M2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional block, and a spatial phase-coded gradient is used to encode the position in the layer direction. Two adjacent regions to be imaged are alternately excited, using pulses with intensities K in three directions. x ,K y ,K z The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; module M2.2 is a motion decoding module;

[0042] Module M2.3: If full sampling is performed, the spatial phase encoding gradient in the motion decoding module is changed and applied repeatedly N times, where N is the number of target layers required, until the same golden angle k spatial radial trajectory line in all target layers is collected. In the N repeated acquisitions, the spatial phase encoding gradient is set to first the center and then the two sides.

[0043] If undersampling is performed, then... Second collection.

[0044] Preferably, module M2.4: repeats modules M2.1 to M2.3, changing the acquired golden angle k-space radial trajectory line in each repetition;

[0045] If full sampling is performed, modules M2.1 to M2.3 will be repeated M times, where M is related to the resolution; if the target resolution is m×m, then

[0046] If undersampling is performed, then modules M2.1 through M2.3 will be repeated. In this case, R represents the undersampling factor;

[0047] Module M2.5: Uses Hadamard encoding to record fluctuation information in different directions; repeats module M2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are:

[0048] D1 = [+1, -1, -1]

[0049] D2 = [+1,+1,+1]

[0050] D3 = [-1, -1, +1]

[0051] D4 = [-1, +1, -1]

[0052] The order is not limited, where +1 represents positive and -1 represents negative.

[0053] Preferably, in module M3:

[0054] Module M3.1: Arranges the raw data of the two adjacent regions that need to be imaged according to their true k-space positions, and fills the missing positions due to undersampling with 0;

[0055] Module M3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image:

[0056]

[0057] Where x is the target fluctuation image to be recovered, y is the acquired raw data, S is the coil sensitivity map, and F... u It is the Fourier transform and undersampling operator. It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform, and λ1 and λ2 are the weights that control the two regularization constraints, respectively.

[0058] Preferably, module M3.3: uses the alternating direction multiplier method to iteratively optimize equation (1) until the objective function converges, obtaining the reconstructed complex image x of the two blocks. slab1 With x slab2 Calculate the phase φ of the complex image respectively, and obtain φ slab1 With φ slab2 ;

[0059] Module M3.4: Decodes the phase image φ, which is caused by the superposition of multi-directional information due to Hadamard encoding, using equation (2):

[0060]

[0061]

[0062] Where, * represents dot product, indicating multiplication at corresponding positions; φ0 is the background phase outside the target wave pattern; E is the Hadamard coding matrix, which is related to the applied gradient direction and is constructed based on the actual direction; φ1, φ2, φ3, and φ4 are the phase patterns obtained by gradient coding using four directions: D1, D2, D3, and D4, respectively; the resulting u = [u x ,u y ,u z ] T For the target fluctuation image.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] 1. This invention uses three-dimensional excitation and acquisition, which improves the intensity of the acquired signal itself, thereby making up for the shortcomings of the original method;

[0065] 2. The present invention uses a multi-block alternating acquisition method, which can maintain the signal-to-noise ratio of the image under short TR conditions, thereby increasing the number of samples within the same vibration cycle and improving acquisition efficiency;

[0066] 3. The acquisition sequence in this invention uses three-dimensional acquisition of the golden angle star-shaped stacked trajectory. Each k-space passes through the center of the k-space of each layer, which is not sensitive to motion and can easily achieve a higher acceleration factor. At the same time, this invention performs under-sampling in both the layer direction and the layer, which, combined with the fast reconstruction algorithm in this invention, makes full use of the three-dimensional spatial and temporal information of the image, and can achieve an acceleration of about 8 times in terms of reconstruction algorithm.

[0067] 4. This invention uses Hadamard encoding, which reduces the number of scans from 6 to 4, and increases the scanning speed to 1.5 times that of traditional encoding methods. Attached Figure Description

[0068] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0069] Figure 1 This is a schematic diagram of the acquisition sequence;

[0070] Figure 2 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0071] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0072] Example 1:

[0073] This invention improves the image signal-to-noise ratio by using three-dimensional acquisition based on the golden angle star-shaped stacked trajectory; it improves displacement coding efficiency and shortens the elastic imaging acquisition time by using a multi-layer block alternating acquisition mechanism; it reduces the number of times multi-directional displacement recording is required in traditional elastic imaging by using the Hadamard displacement coding method, thus shortening the acquisition time; and it achieves rapid sampling and image reconstruction of elastic imaging by applying local low-rank constraints in the three-dimensional spatial domain and sparse transformation in the temporal domain.

[0074] A three-dimensional fast magnetic resonance elastography acquisition and reconstruction method provided by the present invention, such as... Figures 1-2 As shown, it includes:

[0075] Step S1: Use an external vibrator to generate vibrations and transmit them into the human tissue, creating fluctuations within the tissue.

[0076] Step S2: Instruct the magnetic resonance scanner to receive a synchronization signal and apply an image acquisition sequence to record the fluctuation information of human tissue;

[0077] Specifically, in step S2:

[0078] Step S2.1: At the beginning of the sequence, apply an electric field with intensities K in three directions. x ,K y ,K z The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton;

[0079] Step S2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional block, and the position of the layer direction is encoded using a spatial phase-coded gradient. Two adjacent regions to be imaged are alternately excited, using pulses with intensities K in three directions. x ,K y ,K z The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; step S2.2 is a motion decoding module;

[0080] Step S2.3: If full sampling is performed, change the spatial phase encoding gradient in the motion decoding module and repeat it N times, where N is the number of target layers required, until the same golden angle k spatial radial trajectory line in all target layers is collected. In the N repeated collections, the spatial phase encoding gradient is set to first the center and then the two sides.

[0081] If undersampling is performed, then... Second collection.

[0082] Specifically, step S2.4: Repeat steps S2.1 to S2.3, changing the collected golden angle k-space radial trajectory line in each repetition;

[0083] If full sampling is performed, steps S2.1 to S2.3 are repeated M times, where M is related to the resolution; if the target resolution is m×m, then

[0084] If undersampling is performed, repeat steps S2.1 to S2.3. In this case, R represents the undersampling factor;

[0085] Step S2.5: Use Hadamard encoding to record the fluctuation information in different directions; repeat step S2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are:

[0086] D1 = [+1, -1, -1]

[0087] D2 = [+1,+1,+1]

[0088] D3 = [-1, -1, +1]

[0089] D4 = [-1, +1, -1]

[0090] The order is not limited, where +1 represents positive and -1 represents negative.

[0091] Step S3: Obtain the raw data and use a fast reconstruction algorithm to obtain an accurate wave image in human tissue.

[0092] Specifically, in step S3:

[0093] Step S3.1: Arrange the original data of the two adjacent regions that need to be imaged according to their true k-space positions, and fill the missing positions due to undersampling with 0;

[0094] Step S3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image:

[0095]

[0096] Where x is the target fluctuation image to be recovered, y is the acquired raw data, S is the coil sensitivity map, and F... u It is the Fourier transform and undersampling operator. It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform, and λ1 and λ2 are the weights that control the two regularization constraints, respectively.

[0097] Specifically, step S3.3: Iteratively optimize equation (1) using the alternating direction multiplier method until the objective function converges, obtaining the reconstructed complex image x of the two blocks. slab1 With x slab2 Calculate the phase φ of the complex image respectively, and obtain φ slab1 With φ slab2 ;

[0098] Step S3.4: Decode the phase image φ, which is caused by the superposition of multi-directional information due to Hadamard encoding, using equation (2):

[0099]

[0100]

[0101] Where, * represents dot product, indicating multiplication at corresponding positions; φ0 is the background phase outside the target wave pattern; E is the Hadamard coding matrix, which is related to the applied gradient direction and is constructed based on the actual direction; φ1, φ2, φ3, and φ4 are the phase patterns obtained by gradient coding using four directions: D1, D2, D3, and D4, respectively; the resulting u = [u x ,u y ,u z ] T For the target fluctuation image.

[0102] Example 2:

[0103] Example 2 is a preferred embodiment of Example 1, and is used to illustrate the present invention in more detail.

[0104] The present invention also provides a three-dimensional fast magnetic resonance elastography acquisition and reconstruction system. The three-dimensional fast magnetic resonance elastography acquisition and reconstruction system can be implemented by executing the process steps of the three-dimensional fast magnetic resonance elastography acquisition and reconstruction method. That is, those skilled in the art can understand the three-dimensional fast magnetic resonance elastography acquisition and reconstruction method as a preferred embodiment of the three-dimensional fast magnetic resonance elastography acquisition and reconstruction system.

[0105] A three-dimensional fast magnetic resonance elastography acquisition and reconstruction system according to the present invention includes:

[0106] Module M1: Uses an external vibrator to generate vibrations and transmits them into human tissue, creating ripples within the tissue.

[0107] Module M2: Enables the magnetic resonance scanner to receive a synchronization signal and applies an image acquisition sequence to record the wave information of human tissue;

[0108] Specifically, in module M2:

[0109] Module M2.1: At the beginning of the sequence, apply an electric field with intensities K in three directions. x ,K y ,K z The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton;

[0110] Module M2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional block, and a spatial phase-coded gradient is used to encode the position in the layer direction. Two adjacent regions to be imaged are alternately excited, using pulses with intensities K in three directions. x ,K y ,K z The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; module M2.2 is a motion decoding module;

[0111] Module M2.3: If full sampling is performed, the spatial phase encoding gradient in the motion decoding module is changed and applied repeatedly N times, where N is the number of target layers required, until the same golden angle k spatial radial trajectory line in all target layers is collected. In the N repeated acquisitions, the spatial phase encoding gradient is set to first the center and then the two sides.

[0112] If undersampling is performed, then... Second collection.

[0113] Specifically, module M2.4: repeat modules M2.1 to M2.3, changing the acquired golden angle k-space radial trajectory line in each repetition;

[0114] If full sampling is performed, modules M2.1 to M2.3 will be repeated M times, where M is related to the resolution; if the target resolution is m×m, then

[0115] If undersampling is performed, then modules M2.1 through M2.3 will be repeated. In this case, R represents the undersampling factor;

[0116] Module M2.5: Uses Hadamard encoding to record fluctuation information in different directions; repeats module M2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are:

[0117] D1 = [+1, -1, -1]

[0118] D2 = [+1,+1,+1]

[0119] D3 = [-1, -1, +1]

[0120] D4 = [-1, +1, -1]

[0121] The order is not limited, where +1 represents positive and -1 represents negative.

[0122] Module M3: Acquires raw data and uses a fast reconstruction algorithm to obtain accurate wave images of human tissue.

[0123] Specifically, in module M3:

[0124] Module M3.1: Arranges the raw data of the two adjacent regions that need to be imaged according to their true k-space positions, and fills the missing positions due to undersampling with 0;

[0125] Module M3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image:

[0126]

[0127] Where x is the target fluctuation image to be recovered, y is the acquired raw data, S is the coil sensitivity map, and F... uIt is the Fourier transform and undersampling operator. It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform, and λ1 and λ2 are the weights that control the two regularization constraints, respectively.

[0128] Specifically, module M3.3 uses the alternating direction multiplier method to iteratively optimize equation (1) until the objective function converges, obtaining the reconstructed complex image x of the two blocks. slab1 With x slab2 Calculate the phase φ of the complex image respectively, and obtain φ slab1 With φ slab2 ;

[0129] Module M3.4: Decodes the phase image φ, which is caused by the superposition of multi-directional information due to Hadamard encoding, using equation (2):

[0130]

[0131]

[0132] Where, * represents dot product, indicating multiplication at corresponding positions; φ0 is the background phase outside the target wave pattern; E is the Hadamard coding matrix, which is related to the applied gradient direction and is constructed based on the actual direction; φ1, φ2, φ3, and φ4 are the phase patterns obtained by gradient coding using four directions: D1, D2, D3, and D4, respectively; the resulting u = [u x ,u y ,u z ] T For the target fluctuation image.

[0133] Example 3:

[0134] Example 3 is a preferred example of Example 1, and is used to illustrate the present invention in more detail.

[0135] A three-dimensional fast magnetic resonance elastography sequence and reconstruction technique based on alternating multi-block acquisition and Hadamard encoding includes:

[0136] Step 1: Use an external vibrator to generate vibrations and transmit them into the human tissue, creating fluctuations within the tissue.

[0137] Step 2: The magnetic resonance scanner receives the synchronization signal and applies an image acquisition sequence to record the fluctuation information of human tissue.

[0138] Step 3: Obtain the raw data and use a fast reconstruction algorithm to obtain accurate wave images in human tissue.

[0139] Step 2 includes the following steps:

[0140] Step 2.1: At the beginning of the sequence, apply an electric field with intensities [K] in three directions. x ,K y ,K z The Motion Encoding Block (MEG) is used to record the initial position of the vibration on the longitudinal magnetization vector of the proton.

[0141] Step 2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite a three-dimensional slab, and spatial phase encoding (SPE) is used to encode the position in the slab direction. In this step, two slabs are excited alternately; two slabs refer to any two adjacent regions on the human body that need to be imaged. Taking the human brain as an example, if a whole-brain image needs to be acquired, the upper half of the brain is excited first in this sequence to acquire partial data, and then the lower half of the brain is excited to acquire partial data. By repeating this alternating acquisition process, the image of the entire brain is acquired. The same principle applies to the human abdomen and other parts. A pulse with intensities [K] in three directions is used. x ,K y ,K z The radial acquisition sequence of the Motion Decoding Gradient (MDG) is used to acquire the same k-space trajectory (Spoke) in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired. Figure 1 The image shows four time points captured from two layers. Step 2.2 This process is called a Motion Decoding Block.

[0142] Step 2.3: If full sampling is performed, change the spatial phase encoding gradient in the motion decoding module and repeat it N times (N is the required number of target layers) until the same golden angle k spatial radial trajectory line in all target layers is acquired. To avoid the amplitude modulation effect caused by signal attenuation, in the N repeated acquisitions in this step, the spatial phase encoding gradient should be set to first the center and then the sides. If undersampling is performed, then... Second collection.

[0143] Step 2.4: Repeat steps 2.1 to 2.3, changing the acquired golden angle k-space radial trajectory line in each repetition. If full sampling is performed, repeat this process M times, where M depends on the required resolution. If the target resolution is m×m, then... If undersampling is performed, repeat this process. Here, R represents the undersampling factor.

[0144] Step 2.5: Use Hadamard encoding to record fluctuation information in different directions. In this step, repeat step 2.4 four times, changing the gradient directions of MEG and MDG in each repetition. The gradient directions in the four repetitions are D1 = [+1, -1, -1], D2 = [+1, +1, +1], D3 = [-1, -1, +1], and D4 = [-1, +1, -1], in any order. Here, +1 represents the positive direction and -1 represents the negative direction.

[0145] Step 3 includes the following steps:

[0146] Step 3.1: Arrange the original data of the two layers obtained separately according to the actual k-space position, and fill the missing positions caused by undersampling with 0.

[0147] Step 3.2: Using the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the time domain, the undersampled data of the two blocks are reconstructed using compressed sensing and parallel imaging techniques. The following equation (1) is used as the objective function to recover the accurate target wave image.

[0148]

[0149] Where x is the target fluctuation image to be recovered, y is the acquired raw data, S is the coil sensitivity map, and F... u It is the Fourier transform and undersampling operator. It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform, and λ1 and λ2 are the weights that control the two regularization constraints, respectively.

[0150] Step 3.3: Iteratively optimize equation (1) using the alternating direction multiplier method until the objective function converges, obtaining the reconstructed complex image x of the two blocks. slab1 With x slab2 By calculating the phase φ of the complex image separately, we can obtain φ. slab1 With φ slab2 .

[0151] Step 3.4: Process the phase image φ (including φ) resulting from the superposition of multi-directional information due to Hadamard encoding. slab1 With φ slab2 ), and use equation (2) for decoding.

[0152]

[0153]

[0154] Where * represents dot product, indicating multiplication at corresponding positions, and φ0 is the background phase outside the target wave pattern. E is the Hadamard coding matrix, which is related to the gradient direction applied in step 2.5 and needs to be constructed based on the actual direction in step 2.5. φ1, φ2, φ3, and φ4 are the phase patterns obtained by gradient coding in the four directions D1, D2, D3, and D4 in step 2.5, respectively. The resulting u = [u x ,u y ,u z ] T This is the target fluctuation image.

[0155] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0156] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A three-dimensional fast magnetic resonance elastography acquisition and reconstruction method, characterized in that, include: Step S1: Use an external vibrator to generate vibrations and transmit them to human tissues, creating ripples within the tissues; Step S2: Instruct the magnetic resonance scanner to receive a synchronization signal and apply an image acquisition sequence to record the fluctuation information of human tissue; Step S3: Obtain the raw data and use a fast reconstruction algorithm to obtain an accurate wave image in human tissue; In step S2: Step S2.1: At the beginning of the sequence, apply an electric field with intensities in three directions as follows: The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton; Step S2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional block, and the position of the layer direction is encoded using a spatial phase-coded gradient. Two adjacent regions to be imaged are alternately excited, using pulses with intensities in three directions. The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; step S2.2 is a motion decoding module; Step S2.3: If full sampling is performed, change the spatial phase encoding gradient in the motion decoding module and apply it repeatedly. Next, among them For the required number of target layers, continue until the same golden angle k-space radial trajectory line is collected in all target layers. In the repeated acquisition, the spatial phase encoding gradient is set to first the center and then the two sides; If undersampling is performed, then... Secondary collection; In step S3: Step S3.1: Arrange the original data of the two adjacent regions that need to be imaged according to their true k-space positions, and fill the missing positions caused by undersampling with 0; Step S3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image: in, It is necessary to recover an accurate image of the target fluctuation. It is the raw data collected. It is a coil sensitivity diagram. It is the Fourier transform and undersampling operator; It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform. and These are the weights that control the two regularization constraints.

2. The three-dimensional fast magnetic resonance elastography acquisition and reconstruction method according to claim 1, characterized in that, In step S2: Step S2.4: Repeat steps S2.1 to S2.3, changing the collected golden angle k-space radial trajectory line in each repetition; If full sampling is performed, steps S2.1 to S2.3 are repeated. Second-rate, It is related to the resolution; if the target resolution is ,but ; If undersampling is performed, repeat steps S2.1 to S2.

3. Second-rate, This is the undersampling factor; Step S2.5: Use Hadamard encoding to record the fluctuation information in different directions; repeat step S2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are: The order is not limited, among which, Represents a positive direction. It represents the opposite.

3. The three-dimensional fast magnetic resonance elastography acquisition and reconstruction method according to claim 1, characterized in that, In step S3: Step S3.3: Iteratively optimize equation (1) using the alternating direction multiplier method until the objective function converges, obtaining the reconstructed complex images of the two blocks. and ; Calculate the phase of the complex image respectively ,get and ; Step S3.4: Process the phase image with multi-directional information superposition caused by Hadamard encoding. Decoding is performed using equation (2): in, Dot product means multiplying corresponding elements. The background phase outside the target fluctuation chart; The Hadamard encoding matrix is ​​related to the direction in which the gradient is applied and is constructed based on the actual direction. respectively using Phase maps obtained from gradient encoding in four directions; results For the target fluctuation image.

4. A three-dimensional fast magnetic resonance elastography acquisition and reconstruction system, characterized in that, include: Module M1: Uses an external vibrator to generate vibrations and transmits them to human tissues, creating ripples within the tissues. Module M2: Enables the magnetic resonance scanner to receive a synchronization signal and applies an image acquisition sequence to record the wave information of human tissue; Module M3: Acquires raw data and uses a fast reconstruction algorithm to obtain accurate wave images in human tissue; In module M2: Module M2.1: At the beginning of the sequence, apply an electric field with intensities in three directions as follows: The motion encoding module of the displacement encoding gradient records the initial position of the vibration on the longitudinal magnetization vector of the proton; Module M2.2: In the subsequent imaging module, a wideband minimum-phase radio frequency pulse is used to excite the three-dimensional layer block, and a spatial phase-coded gradient is used to encode the position of the layer direction. Two adjacent regions to be imaged are alternately excited, using pulses with intensities in three directions. The radial acquisition sequence of the displacement decoding gradient is used to acquire the same k-space trajectory line in the same spatial phase encoding at different time points until the required t fluctuation time points are acquired; module M2.2 is a motion decoding module; Module M2.3: If full sampling is performed, the spatial phase encoding gradient in the motion decoding module is varied and repeatedly applied. Next, among them For the required number of target layers, continue until the same golden angle k-space radial trajectory line is collected in all target layers. In the repeated acquisition, the spatial phase encoding gradient is set to first the center and then the two sides; If undersampling is performed, then... Secondary collection; In module M3: Module M3.1: Arranges the raw data of the two adjacent regions that need to be imaged according to their true k-space positions, and fills the missing positions due to undersampling with 0; Module M3.2: Utilizing the local low-rank constraint in the three-dimensional spatial domain and the sparse transformation in the temporal domain, compressed sensing and parallel imaging techniques are used to reconstruct the undersampled data of the two blocks respectively. The following equation (1) is used as the objective function to recover the accurate target fluctuation image: in, It is necessary to recover the target fluctuation image. It is the raw data collected. It is a coil sensitivity diagram. It is the Fourier transform and undersampling operator; It is a local low-rank constraint in three-dimensional space. It is a time-domain wavelet transform. and These are the weights that control the two regularization constraints.

5. The three-dimensional fast magnetic resonance elastography acquisition and reconstruction system according to claim 4, characterized in that, In module M2: Module M2.4: Repeat modules M2.1 to M2.3, changing the acquired golden angle k-space radial trajectory line in each repetition; If full sampling is performed, modules M2.1 through M2.3 will be repeated. Second-rate, It is related to the resolution; if the target resolution is ,but ; If undersampling is performed, then modules M2.1 through M2.3 will be repeated. Second-rate, This is the undersampling factor; Module M2.5: Uses Hadamard encoding to record fluctuation information in different directions; repeats module M2.4 four times, changing the gradient direction of the displacement encoding gradient and the displacement decoding gradient in each repetition; the gradient directions in the four repetitions are: The order is not limited, among which, Represents a positive direction. It represents the opposite.

6. The three-dimensional fast magnetic resonance elastography acquisition and reconstruction system according to claim 4, characterized in that, In module M3: Module M3.3: Iteratively optimize equation (1) using the alternating direction multiplier method until the objective function converges, obtaining the reconstructed complex images of the two blocks. and ; Calculate the phase of the complex image respectively ,get and ; Module M3.4: Phase images with multi-directional information superposition caused by Hadamard encoding Decoding is performed using equation (2): in, Dot product means multiplying corresponding elements. The background phase outside the target fluctuation chart; The Hadamard encoding matrix is ​​related to the direction in which the gradient is applied and is constructed based on the actual direction. respectively using Phase maps obtained from gradient encoding in four directions; results For the target fluctuation image.

Images