An Optimization Method for Quantitative T2 Imaging Acquisition Based on Singular Value Entropy Multioverlap Echo Separation

Abstract

The application discloses a method for optimizing a multiple overlapping echo decomposition (MOLED) T2 imaging sequence based on singular value entropy. The method comprises the following steps: acquiring a MOLED-T2 imaging sequence; determining sampling parameters; adding the MOLED-T2 imaging sequence to a magnetic resonance imaging platform, setting the sampling parameters, and acquiring MOLED images through data acquisition; establishing an optimization objective function of the MOLED-T2 imaging sequence based on a singular value entropy formula, echo shift gradient setting and the MOLED images; calculating an optimal solution of the optimization objective function with the sampling parameters as constraint conditions to obtain an optimized MOLED-T2 imaging sequence; acquiring MOLED images for training and reconstruction; training a deep neural network; and inputting the MOLED images for reconstruction into the trained deep neural network model to reconstruct a T2 parameter map. The method optimizes echo shift gradient setting of the MOLED-T2 imaging sequence, increases information acquisition amount, and improves the quality of the T2 parameter map reconstructed by the deep neural network.

Description

Technical Field

[0001] This invention relates to the field of magnetic resonance imaging technology, and in particular to an optimized method for quantitative T2 imaging acquisition based on singular value entropy and multi-overlapping echo separation. Background Technology

[0002] Multiple Overlapping-Echo Detachment (MOLED-T2) quantitative T2 imaging sequences can acquire images composed of overlapping echo signals with different phase evolutions and T2 weighting in a single scan. Deep neural networks can then reconstruct quantitative T2 parameter maps from the acquired MOLED images. The MOLED-T2 imaging sequence uses echo shift gradient settings to deviate the echo signals from the k-space center in both the phase and frequency encoding dimensions, providing a basis for deep neural network reconstruction of the T2 parameter maps. The echo shift gradient setting scheme affects the overlap of the echo signals, thus affecting the amount of information contained in the MOLED image, and consequently, the quality of the T2 parameter map reconstructed by the deep neural network from the MOLED image. However, currently, the echo shift gradient setting in MOLED-T2 imaging sequences mainly relies on operator experience, lacking optimized methods.

[0003] Singular value entropy reflects the complexity and information content of the singular value distribution of a two-dimensional matrix. When the singular value distribution is uniform, the singular value entropy is large, indicating that the matrix contains relatively dispersed and rich information. Conversely, when the singular value distribution is concentrated in a few large singular values, the singular value entropy is small, indicating that the matrix's information is concentrated in certain main patterns. Singular value entropy can be used to calculate the information content of MOLED images, thus quantitatively measuring the amount of information acquired in MOLED-T2 imaging sequences. Therefore, it can be used to optimize echo shift gradient settings and improve the accuracy, precision, and resolution of T2 parameter map reconstruction by deep neural networks. Summary of the Invention

[0004] The main objective of this invention is to propose an optimization method for quantitative T2 imaging acquisition based on singular value entropy using multi-overlapping echo separation, in order to optimize the echo shift gradient setting scheme of MOLED-T2 imaging sequence, increase the amount of information acquired in MOLED-T2 imaging sequence, and improve the quality of T2 parameter map reconstructed by deep neural network.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] An optimized acquisition method for quantitative T2 imaging of magnetic resonance imaging based on singular value entropy and multi-overlapping echo separation includes the following steps:

[0007] S1 acquires the MOLED-T2 imaging sequence, which is the multi-overlap echo separation magnetic resonance quantitative T2 imaging sequence.

[0008] S2 determines the sampling parameters of the MOLED-T2 imaging sequence, including echo shift gradient settings and other sampling parameters;

[0009] S3 adds an MOLED-T2 imaging sequence to the magnetic resonance imaging platform, sets sampling parameters, and acquires data to obtain MOLED images;

[0010] S4 establishes an optimization objective function for the MOLED-T2 imaging sequence based on the singular value entropy formula, echo shift gradient setting, and MOLED images;

[0011] S5 uses the sampling parameters as constraints to calculate the optimal solution of the objective function and obtain the optimized MOLED-T2 imaging sequence;

[0012] S6 uses an optimized MOLED-T2 imaging sequence to acquire data on a magnetic resonance imaging platform, obtaining optimized MOLED images, which are divided into training MOLED images and reconstruction MOLED images.

[0013] S7 uses training MOLED images to train a deep neural network, resulting in a trained deep neural network model.

[0014] S8 will reconstruct the T2 parameter map by inputting the MOLED image into the trained deep neural network model.

[0015] In step S1, the MOLED-T2 imaging sequence is obtained, and the specific structure of the sequence is as follows:

[0016] α1,G1,α2,G2,…,α N G N β, sampling echo train where α i (i = 1, 2, 3, ..., N) are the excitation pulses, N represents the number of excitation pulses, and N excitation pulses produce N main echoes; G i β represents the echo shift gradient, and each echo shift gradient includes a frequency encoding direction shift gradient and a phase encoding direction shift gradient; β is the refocusing pulse; the sampled echo chain consists of a series of positive and negative gradients in the frequency encoding direction and a series of equal "blips" gradients in the phase encoding direction.

[0017] In step S2, the sampling parameters of the MOLED-T2 sequence are determined, including the echo shift gradient setting and other sampling parameters, specifically including:

[0018] Determine the MOLED-T2 imaging sequence parameters, including the size and shape of the excitation pulse, the size and duration of the echo shift gradient, the size and shape of the refocusing pulse, the echo time, the sampling echo train length, and the echo interval, etc.

[0019] Determine parameters such as imaging field of view and imaging matrix.

[0020] In step S3, an MOLED-T2 imaging sequence is added to the magnetic resonance imaging platform, sampling parameters are set, and data is acquired to obtain an MOLED image. Specifically, this includes:

[0021] Input the MOLED-T2 imaging sequence into the magnetic resonance imaging simulation platform;

[0022] Obtain simulation templates with a set amount of parameters. Each simulation template includes a T1 parameter map, a T2 parameter map, and a proton density map. Simulation templates can be obtained from public datasets or generated by simulation.

[0023] By setting sampling parameters and using a simulation template as the imaging object, a simulated scan is performed on a magnetic resonance simulation platform to obtain the MOLED imaging signal.

[0024] The MOLED imaging signal is rearranged into a two-dimensional k-space signal, and then an inverse Fourier transform is performed to obtain the MOLED image.

[0025] In step S4, an optimization objective function for the MOLED-T2 imaging sequence is established based on the singular value entropy formula, echo shift gradient setting, and MOLED image, as follows:

[0026] The information content f(G) of an MOLED image is expressed as:

[0027]

[0028] The following optimization objective function is then established:

[0029] max:f(G)

[0030] Where G represents the echo shift gradient setting of the MOLED-T2 imaging sequence, which is the parameter to be optimized; I(G) represents the complex two-dimensional matrix of the MOLED image obtained by simulated scanning of the MOLED-T2 imaging sequence under the echo shift gradient setting; t represents the t-th simulation template; Q represents the total number of simulation templates; and SvdEn(·) represents the amount of information in a single MOLED image calculated using the singular value entropy formula.

[0031] The information content of a single MOLED image is calculated using the singular value entropy formula. The specific process is as follows:

[0032] The singular value matrix Σ is obtained using the singular value decomposition formula:

[0033] I=UΣV T

[0034] Wherein, the Σ matrix has a size of H×W, where H and W represent the number of pixels in the frequency-coded and phase-coded dimensions of the MOLED image, respectively, and the magnitude of the singular values ​​reflects the structure and detail information of the MOLED image; I is a complex two-dimensional matrix of the MOLED image with a size of H×W; U is an H-order orthogonal matrix, representing the left singular value matrix of I, and its column vectors reflect the projection of matrix I onto the column vectors; V T It is an orthogonal matrix of order W, representing the right singular value matrix of I, whose row vectors reflect the projection of matrix I onto the row vectors; T represents the transpose;

[0035] Extracting the singular values ​​from Σ and arranging them in order of magnitude yields the singular value vector σ of the MOLED image.

[0036] σ=[σ1,σ2,σ3,…,σ L ]

[0037] L represents the larger of H and W; the singular value vector σ is transformed into a probability density distribution vector p:

[0038]

[0039] σ j (j=1,2,3,…,L) represents the j-th element of the singular value vector σ; the entropy of the probability density distribution vector p is calculated using the singular value entropy formula, which is used to measure the information content of the MOLED image. The calculation formula is as follows:

[0040]

[0041] Where, p j It represents the j-th element of the probability density distribution vector p; SvdEn is the singular value entropy, which reflects the information content of the MOLED image. The larger the singular value entropy, the more information the MOLED image contains.

[0042] In step S5, the sampling parameters are used as constraints, as detailed below:

[0043] The position of the main echo in k-space is determined by the echo shift gradient. Using the position of the main echo in k-space as a constraint, the constraint can be set using the following formula:

[0044]

[0045] k ROi k PEi G represents the positions of the i-th main echo frequency encoding direction and phase encoding direction in k-space, respectively; ROi G PEiδ represents the shift gradient in the frequency encoding direction and the phase encoding direction of the i-th echo shift gradient, respectively; i Represents the duration of the i-th echo shift gradient; FOV RO FOV PE These represent the imaging field of view in the frequency encoding direction and the phase encoding direction, respectively; γ represents the gyromagnetic ratio of the nucleus.

[0046] In step S5, the optimal solution of the objective function is calculated to obtain the optimized MOLED-T2 imaging sequence. The specific process is as follows:

[0047] First, under constraints, keeping other sampling parameters constant, set the echo shift gradient of P groups of MOLED-T2 imaging sequences; use the magnetic resonance imaging platform to generate P groups of MOLED images with corresponding echo shift gradient settings, and calculate the information content of the P groups of MOLED images; each group contains Q MOLED images.

[0048] Secondly, the optimal solution of the objective function is selected based on the information content of the P group of MOLED images. The optimal echo shift gradient setting of the MOLED-T2 imaging sequence is determined based on the optimal solution. The MOLED-T2 imaging sequence under this echo shift gradient setting is the optimized MOLED-T2 imaging sequence.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] This invention provides an optimization method for quantitative T2 imaging acquisition based on singular value entropy using multi-overlap echo separation: acquiring a MOLED-T2 imaging sequence; determining the sampling parameters of the MOLED-T2 imaging sequence; adding the MOLED-T2 imaging sequence to the magnetic resonance imaging platform, setting the sampling parameters, and acquiring data to obtain MOLED images; establishing an optimization objective function for the MOLED-T2 imaging sequence based on the singular value entropy formula, echo shift gradient settings, and MOLED images; calculating the optimal solution of the optimization objective function using the sampling parameters as constraints to obtain an optimized MOLED-T2 imaging sequence; obtaining optimized MOLED images based on the optimized MOLED-T2 imaging sequence; training a deep neural network using the optimized MOLED images to obtain a trained deep neural network model; and reconstructing the T2 parameter map using the optimized MOLED images and the trained deep neural network model.

[0051] This invention can optimize the echo shift gradient setting scheme of MOLED-T2 imaging sequences, increase the amount of information collected, and improve the quality of T2 parameter maps reconstructed by deep neural networks. Attached Figure Description

[0052] Figure 1This is a flowchart of the optimization method for quantitative T2 imaging acquisition based on singular value entropy using multi-overlapping echo separation magnetic resonance in an embodiment of the present invention;

[0053] Figure 2 This is a structural diagram of the MOLED-T2 imaging sequence in an embodiment of the present invention;

[0054] Figure 3 The images shown are k-space data (a) and MOLED image (b) acquired by the MOLED-T2 imaging sequence in this embodiment of the invention, k-space data (c) and MOLED image (d) acquired by the optimized MOLED-T2 imaging sequence, and T2 parameter map (e) reconstructed by the deep learning network. Detailed Implementation

[0055] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] See Figure 1 An optimized acquisition method for quantitative T2-weighted magnetic resonance imaging based on singular value entropy and multi-overlapping echo separation includes the following steps:

[0057] S1 acquires the MOLED-T2 imaging sequence. The MOLED-T2 imaging sequence is the multi-overlap echo-separated magnetic resonance quantitative T2 imaging sequence.

[0058] In practical applications, the MOLED-T2 imaging sequence includes four excitation pulses α i (i = 1, 2, 3, 4), 4 excitation pulses generate 4 main echoes; refocusing pulse β; an echo shift gradient is applied after each excitation pulse, and each echo shift gradient G i Includes frequency coding direction shift gradient G ROi and phase encoding direction shift gradient G PEi The sampling echo chain consists of a series of positive and negative gradients in the frequency encoding direction and a series of equal "blips" gradients in the phase encoding direction.

[0059] Figure 2 A structural diagram showing the MOLED-T2 imaging sequence.

[0060] S2 determines the sampling parameters of the MOLED-T2 imaging sequence, including echo shift gradient settings and other sampling parameters.

[0061] In practical applications, the sampling parameters are set as follows:

[0062] All four excitation pulses are 30° sinc pulses;

[0063] The refocusing pulse is a 180° sinc pulse;

[0064] The four echo times are TE1 = 22ms, TE2 = 52ms, TE3 = 82ms, and TE4 = 110ms.

[0065] The sampling echo train length is 128, and the echo interval is 0.465ms;

[0066] Assuming the total area of ​​the gradients in both the frequency encoding and phase encoding directions of the sampled echo chain is 1 unit area, and the initial echo shift gradient is set to (G RO1 G PE1 ) = (-0.19, -0.18), (G RO2 G PE2 ) = (0.10, -0.24), (G RO3 G PE3 ) = (-0.18, -0.25), (G RO4 G PE4 = (-0.38, -0.13);

[0067] The field of view (FOV) is 220mm × 220mm, and the imaging matrix is ​​128 × 128.

[0068] S3 adds an MOLED-T2 imaging sequence to the magnetic resonance imaging platform, sets sampling parameters, and acquires data to obtain MOLED images;

[0069] In practical applications, the MOLED-T2 sequence is input into the MRiLab simulation platform;

[0070] The T1 parameter map was fixed as a constant map for 4000 ms; proton density maps and T2 parameter maps were generated using multi-contrast images from the IXI public dataset. A total of 5000 simulation templates were obtained.

[0071] A simulated template was used as the imaging object to perform a simulated scan on a magnetic resonance simulation platform to obtain the MOLED imaging signal.

[0072] The MOLED imaging signal is rearranged into a two-dimensional k-space signal, and then an inverse Fourier transform is performed to obtain the MOLED image.

[0073] Each simulation template simulates scanning one MOLED image, resulting in a total of 5000 MOLED images. Figure 3 Figures (a) and (b) show a representative two-dimensional k-space data and its MOLED image obtained by simulated scanning of the MOLED-T2 imaging sequence in this embodiment.

[0074] S4 establishes an optimization objective function for the MOLED-T2 imaging sequence based on the singular value entropy formula, echo shift gradient setting, and MOLED images.

[0075] In practical applications, the amount of MOLED-T2 imaging sequence information acquired, f(G), is expressed as:

[0076]

[0077] The following optimization objective function is then established:

[0078] max:f(G)

[0079] Where G represents the echo shift gradient setting of the MOLED-T2 imaging sequence, which is the parameter to be optimized; I(G) represents the complex two-dimensional matrix of the MOLED image obtained by simulated scanning of the MOLED-T2 imaging sequence under this echo shift gradient setting; t represents the t-th simulation template; Q represents the total number of simulation templates, which is 5000; and SvdEn(·) represents the amount of information in a single MOLED image calculated using the singular value entropy formula.

[0080] S5 uses the sampling parameters as constraints to calculate the optimal solution of the objective function and obtain the optimized MOLED-T2 imaging sequence.

[0081] In practical applications, the position of the main echo in k-space is determined by the echo shift gradient. Using the position of the main echo in k-space as a constraint, the constraint can be set using the following formula:

[0082]

[0083] k ROi k PEi G represents the positions of the i-th main echo frequency encoding direction and phase encoding direction in k-space, respectively, in units of 1 / mm; ROi G PEi δ represents the shift gradient of the i-th echo in the frequency encoding direction and the phase encoding direction, respectively; i γ represents the duration of the i-th echo shift gradient; γ represents the gyromagnetic ratio of the nucleus.

[0084] The process of calculating the optimal solution of the objective function and obtaining the optimized MOLED-T2 imaging sequence is as follows:

[0085] First, under constraints, keeping other sampling parameters constant, 100 sets of echo shift gradients for MOLED-T2 imaging sequences were set. Then, using a magnetic resonance imaging platform, 100 sets of MOLED images corresponding to the echo shift gradient settings were generated, and the information content of each set of MOLED images was calculated. Each set of MOLED images includes 5000 MOLED images.

[0086] Secondly, the optimal solution of the objective function is selected based on the information content of 100 MOLED images. The optimal echo shift gradient setting for the MOLED-T2 imaging sequence is then determined based on this optimal solution. The MOLED-T2 imaging sequence under this echo shift gradient setting is the optimized MOLED-T2 imaging sequence. The echo shift gradient of the optimized MOLED-T2 imaging sequence is set as (G... RO1 G PE1 ) = (-0.36, -0.18), (G RO2 G PE2 )=(0.19,-0.24),(G RO3 G PE3 ) = (-0.24, -0.25), (G RO4 G PE4 = (-0.36, -0.13).

[0087] The S6 uses an optimized MOLED-T2 imaging sequence to acquire data on a magnetic resonance imaging platform, obtaining optimized MOLED images, which are divided into training MOLED images and reconstruction MOLED images.

[0088] In practical applications, an optimized MOLED-T2 imaging sequence was used on the MRiLab simulation platform to scan the simulation template and obtain the k-space signal. Considering the variations in the real experimental environment, unstable factors were introduced, including excitation pulse angle deviation, echo shift gradient deviation, and noise. The k-space signal was then converted into an MOLED image using an inverse Fourier transform. The image was normalized with the maximum amplitude of the MOLED image set to 1. A total of 5000 optimized MOLED images were acquired, of which 4500 were used as training images and 500 were used as reconstruction images. Figure 3 Figures (c) and (d) show a representative two-dimensional k-space data and its MOLED image obtained from the simulated scanning of the optimized MOLED-T2 imaging sequence in this embodiment.

[0089] S7 uses training MOLED images to train a deep neural network, resulting in a trained deep neural network model.

[0090] In practical applications, the deep neural network is a 5-layer U-Net, built using the common deep learning framework PyTorch. The reconstruction results are constrained using the L1 norm. A single training MOLED image and the T2 parameter map in the corresponding simulation template are used to construct a training sample, resulting in 4500 training samples, which are then used to train the deep neural network.

[0091] The deep neural network was trained using MOLED images, with the T2 parameter map in the simulation template as the label.

[0092] S8 will reconstruct the T2 parameter map by inputting the MOLED image into the trained deep neural network model. Figure 3 Figure (e) shows a typical T2 parameter plot of a deep neural network reconstruction.

[0093] This invention acquires MOLED-T2 imaging sequences; determines the sampling parameters of the MOLED-T2 imaging sequences; adds the MOLED-T2 imaging sequences to a magnetic resonance imaging platform, sets the sampling parameters, and performs data acquisition to obtain MOLED images; establishes an optimization objective function for the MOLED-T2 imaging sequences based on the singular value entropy formula, echo shift gradient settings, and MOLED images; calculates the optimal solution of the optimization objective function using the sampling parameters as constraints to obtain the optimized MOLED-T2 imaging sequences; acquires MOLED images for training and reconstruction based on the optimized MOLED-T2 imaging sequences; trains a deep neural network using the training MOLED images to obtain a trained deep neural network model; and inputs the reconstruction MOLED images into the trained deep neural network model to reconstruct the T2 parameter map. Experiments show that this invention can optimize the echo shift gradient setting scheme of the MOLED-T2 imaging sequences, increase the amount of information acquired, and improve the quality of the T2 parameter map reconstructed by the deep neural network.

[0094] This specification uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An optimized acquisition method for quantitative T2 imaging of magnetic resonance imaging based on singular value entropy and multi-overlapping echo separation, characterized in that, Includes the following steps: S1 acquires the MOLED-T2 imaging sequence, which is the multi-overlap echo separation magnetic resonance quantitative T2 imaging sequence. S2 determines the sampling parameters of the MOLED-T2 imaging sequence, including echo shift gradient settings and other sampling parameters; S3 adds an MOLED-T2 imaging sequence to the magnetic resonance imaging platform, sets sampling parameters, and acquires data to obtain MOLED images; S4 establishes an optimization objective function for the MOLED-T2 imaging sequence based on the singular value entropy formula, echo shift gradient setting, and MOLED images, as follows: The information content f(G) of an MOLED image is expressed as: The following optimization objective function is then established: max:f(G) Where G represents the echo shift gradient setting of the MOLED-T2 imaging sequence, which is the parameter to be optimized; I(G) represents the complex two-dimensional matrix of the MOLED image obtained by simulated scanning of the MOLED-T2 imaging sequence under the echo shift gradient setting; t represents the t-th simulation template; Q represents the total number of simulation templates; and SvdEn(·) represents the amount of information in a single MOLED image calculated using the singular value entropy formula. The specific process for calculating the information content of a single MOLED image using the singular value entropy formula is as follows: Obtaining the singular value matrix using the singular value decomposition formula : I = UΣV T in, The matrix has a size of H×W, where H and W represent the number of pixels in the frequency-coded and phase-coded dimensions of the MOLED image, respectively. The magnitude of the singular values ​​reflects the structure and detail information of the MOLED image. I is a complex two-dimensional matrix of the MOLED image with a size of H×W. U is an H-order orthogonal matrix, representing the left singular value matrix of I, and its column vectors reflect the projection of matrix I onto the column vectors. V T It is an orthogonal matrix of order W, representing the right singular value matrix of I, whose row vectors reflect the projection of matrix I onto the row vectors; T represents the transpose; extract The singular values ​​in the MOLED image are obtained by arranging them in order of magnitude. : L represents the larger of H and W; the singular value vector σ is transformed into a probability density distribution vector p: σ j (j = 1, 2, 3, …, L) represents the j-th element of the singular value vector σ; the entropy of the probability density distribution vector p is calculated using the singular value entropy formula, which is used to measure the information content of the MOLED image. The calculation formula is as follows: Where, p j This represents the j-th element of the probability density distribution vector p; SvdEn is the singular value entropy, which reflects the information content of the MOLED image. The larger the singular value entropy, the more information the MOLED image contains. S5 uses the sampling parameters as constraints to calculate the optimal solution of the objective function and obtain the optimized MOLED-T2 imaging sequence; S6 uses an optimized MOLED-T2 imaging sequence to acquire data on a magnetic resonance imaging platform, obtaining optimized MOLED images, which are divided into training MOLED images and reconstruction MOLED images. S7 uses training MOLED images to train a deep neural network, resulting in a trained deep neural network model. S8 will reconstruct the T2 parameter map by inputting the MOLED image into the trained deep neural network model.

2. The optimized acquisition method for quantitative T2 imaging based on singular value entropy using multi-overlapping echo separation magnetic resonance as described in claim 1, characterized in that, In step S1, the acquisition of the MOLED-T2 imaging sequence is specifically represented by the following structure: α1,G1,α2,G2,…,α N G N β, sampling echo chain Where α i (i = 1, 2, 3, ..., N) are the excitation pulses, and N represents the number of excitation pulses. N excitation pulses produce N main echoes; G i β represents the echo shift gradient, and each echo shift gradient includes a frequency encoding direction shift gradient and a phase encoding direction shift gradient; β is the refocusing pulse; the sampled echo chain consists of a series of positive and negative gradients in the frequency encoding direction and a series of equal "blips" gradients in the phase encoding direction.

3. The optimized acquisition method for quantitative T2 imaging based on singular value entropy using multi-overlapping echo separation magnetic resonance as described in claim 1, characterized in that, In step S2, determining the sampling parameters of the MOLED-T2 imaging sequence includes echo shift gradient settings and other sampling parameters, specifically including: Determine the MOLED-T2 imaging sequence parameters, including the size and shape of the excitation pulse, the size and duration of the echo shift gradient, the size and shape of the refocusing pulse, the echo time, the sampling echo train length, and the echo interval; Determine the imaging field of view and imaging matrix.

4. The optimized acquisition method for quantitative T2 imaging based on singular value entropy using multi-overlapping echo separation magnetic resonance as described in claim 1, characterized in that, In step S3, adding the MOLED-T2 imaging sequence to the magnetic resonance imaging platform, setting sampling parameters, and acquiring data to obtain MOLED images specifically includes: Input the MOLED-T2 imaging sequence into the magnetic resonance imaging simulation platform; Obtain simulation templates with a set amount of parameters. Each simulation template includes a T1 parameter map, a T2 parameter map, and a proton density map. The simulation templates can be obtained from public datasets or generated by simulation. By setting sampling parameters and using a simulation template as the imaging object, a simulated scan is performed on a magnetic resonance simulation platform to obtain the MOLED imaging signal. The MOLED imaging signal is rearranged into a two-dimensional k-space signal, and then an inverse Fourier transform is performed to obtain the MOLED image.

5. The optimized acquisition method for quantitative T2 imaging based on singular value entropy using multi-overlapping echo separation magnetic resonance as described in claim 1, characterized in that, In step S5, the constraint condition based on the sampling parameters is as follows: The position of the main echo in k-space is determined by the echo shift gradient. The position of the main echo in k-space is used as a constraint condition, which is set by the following formula: k ROi k PEi G represents the positions of the i-th main echo frequency encoding direction and phase encoding direction in k-space, respectively; ROi G PEi δ represents the shift gradient in the frequency encoding direction and the phase encoding direction of the i-th echo shift gradient, respectively; i This represents the duration of the i-th echo shift gradient; FOV RO FOV PE These represent the imaging field of view sizes in the frequency encoding direction and the phase encoding direction, respectively; Represents the gyromagnetic ratio of the nucleus.

6. The optimized acquisition method for quantitative T2 imaging based on singular value entropy using multi-overlapping echo separation magnetic resonance as described in claim 1, characterized in that, In step S5, the optimal solution of the objective function is calculated to obtain the optimized MOLED-T2 imaging sequence. The specific process is as follows: First, under constraints, keeping other sampling parameters constant, set the echo shift gradient of P groups of MOLED-T2 imaging sequences; use the magnetic resonance imaging platform to generate P groups of MOLED images with corresponding echo shift gradient settings, and calculate the information content of the P groups of MOLED images; each group contains Q MOLED images. Secondly, the optimal solution of the objective function is selected based on the information content of the P group of MOLED images. The optimal echo shift gradient setting of the MOLED-T2 imaging sequence is determined based on the optimal solution. The MOLED-T2 imaging sequence under this echo shift gradient setting is the optimized MOLED-T2 imaging sequence.

Images