WO2020119581A1 - 磁共振参数成像方法、装置、设备及存储介质 - Google Patents

磁共振参数成像方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2020119581A1
WO2020119581A1 PCT/CN2019/123448 CN2019123448W WO2020119581A1 WO 2020119581 A1 WO2020119581 A1 WO 2020119581A1 CN 2019123448 W CN2019123448 W CN 2019123448W WO 2020119581 A1 WO2020119581 A1 WO 2020119581A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
parameter
current
intermediate image
compensation coefficient
Prior art date
Application number
PCT/CN2019/123448
Other languages
English (en)
French (fr)
Inventor
朱燕杰
刘元元
梁栋
刘新
郑海荣
Original Assignee
深圳先进技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳先进技术研究院 filed Critical 深圳先进技术研究院
Publication of WO2020119581A1 publication Critical patent/WO2020119581A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/003—Reconstruction from projections, e.g. tomography
    • G06T11/005—Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]

Definitions

  • the present application relates to the field of image processing, for example, to a magnetic resonance parameter imaging method, device, equipment, and storage medium.
  • T 1 ⁇ spin-lattice relaxation in the rotating frame
  • the related technologies either reduce the number of TSL or adopt fast imaging technology. Since the T 1 ⁇ parameter map is determined by the T 1 ⁇ weighted image, then for the former, due to the reduction of TSL, the number of T 1 ⁇ parameter weighted images is reduced, which in turn leads to a reduction in the accuracy of the T 1 ⁇ parameter map, that is, a reduction in quantitative accuracy;
  • the current commercial rapid imaging technology is mainly parallel imaging (such as sensitivity coding (SENSE), generalized automatic calibration part parallel acquisition (GRAPPA), etc.), due to the limitation of parallel imaging array coils, the higher the acceleration factor, the higher The lower the signal-to-noise ratio of the obtained T 1 ⁇ parameter-weighted image, the lower the quality accuracy of the T 1 ⁇ parameter map, that is, the lower the quantitative accuracy, so the scan speed of parallel imaging can usually only reach 2-3 times.
  • Embodiments of the present invention provide a magnetic resonance parametric imaging method, device, equipment, and storage medium to solve the technical problem that it is difficult for the related art to have both imaging speed and imaging quality in the parametric imaging method.
  • An embodiment of the present invention provides a magnetic resonance parameter imaging method, including:
  • a preset Fourier transform is used to convert the magnetic resonance data to an image domain to obtain an initial image, and a current compensation coefficient is determined based on a double exponential relaxation model and the initial image;
  • the initial image is compensated based on the current compensation coefficient to obtain the compensated initial image, and the compensated initial image is input as the first iteration input image to the L+S model (low-rank plus sparse, L+S, that is, low rank plus sparse Model), update the current compensation coefficient according to the current intermediate image iteratively generated during the image reconstruction process of the L+S model, and use the updated current compensation coefficient to compensate the current intermediate image to generate the input after the compensation of the next iteration process
  • the current intermediate image until the iteration converges, and the intermediate image generated by the last iteration is used as the parameter-weighted image;
  • the parameter-weighted image is nonlinearly fitted to obtain a parameter map.
  • An embodiment of the present invention also provides a magnetic resonance parameter imaging device, including:
  • the data acquisition module is set to acquire the magnetic resonance data of the target object in an undermined manner
  • An initial image module configured to convert the magnetic resonance data to an image domain using a preset Fourier transform to obtain an initial image, and determine a current compensation coefficient based on a double exponential relaxation model and the initial image;
  • the parameter-weighted image determination module is set to compensate the initial image based on the current compensation coefficient to obtain the compensated initial image, and input the compensated initial image as the input image of the first iteration to the L+S model, according to the L+S model in the image
  • the current intermediate image is updated by iteratively generating the current compensation coefficient, and the updated current compensation coefficient is used to compensate the current intermediate image to generate the current intermediate image after the input of the next iteration process until the iteration converges, and The intermediate image generated in the last iteration is used as the parameter-weighted image;
  • the parameter map determination module is configured to perform nonlinear fitting on the parameter-weighted image using the double exponential relaxation model to obtain a parameter map.
  • An embodiment of the present invention also provides a magnetic resonance device, including:
  • One or more processors are One or more processors;
  • Storage device for storing one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors implement the magnetic resonance parameter imaging method as described above.
  • Embodiments of the present invention also provide a storage medium containing computer-executable instructions, which when executed by a computer processor are used to perform the magnetic resonance parameter imaging method as described above.
  • the technical solution of the magnetic resonance parameter imaging method can acquire the magnetic resonance data of the target object in an under-acquisition manner, which can increase the scanning speed of the magnetic resonance data, and uses the preset Fourier transform to convert the magnetic resonance data to an image Field to get the initial image, determine the current compensation coefficient based on the double exponential relaxation model and the initial image; compensate the initial image based on the current compensation coefficient to obtain the compensated initial image, and input the compensated initial image as the input image for the first iteration L+S model, according to the L+S model in the image reconstruction process, iteratively generates the current intermediate image by updating the current compensation coefficient, and uses the updated current compensation coefficient to compensate the current intermediate image to generate the input compensation for the next iteration process After the current intermediate image is converged until the iteration, the intermediate image generated by the last iteration is used as the parameter-weighted image; the double-exponential relaxation model is used to nonlinearly fit the parameter-weighted image to obtain the parameter map.
  • the double exponential relaxation model can more accurately represent the change trend of tissues over time, especially for some complex tissues that have different proton components and interact with each other.
  • the exponential model is inaccurate to describe.
  • the T 1 ⁇ s parameter map and T 1 ⁇ l parameter map obtained by the double exponential relaxation model can better describe the interaction between free water protons and bound water protons, and it is more helpful to explore The underlying biophysical mechanism of T 1 ⁇ relaxation.
  • FIG. 1 is a flowchart of a magnetic resonance parameter imaging method according to Embodiment 1 of the present invention
  • FIG. 2 is a structural block diagram of a magnetic resonance parameter imaging apparatus provided in Embodiment 3 of the present invention.
  • Embodiment 3 is a structural block diagram of a magnetic resonance device provided in Embodiment 4 of the present invention.
  • FIG. 1 is a flowchart of a magnetic resonance parameter imaging method provided in Embodiment 1 of the present invention.
  • the technical solution of this embodiment is suitable for the case of quickly acquiring a high-quality parameter map.
  • the method may be performed by the magnetic resonance parameter imaging apparatus provided in the embodiment of the present invention.
  • the apparatus may be implemented in software and/or hardware, and configured to be applied in a processor.
  • the method specifically includes the following steps:
  • this embodiment is based on the sparse sampling theory, and acquires the magnetic resonance data of the target object in an under-recovery mode, specifically: because the parameter image often needs to acquire multiple TSL (spin lock time) T 1 ⁇ parameter weighted image.
  • TSL spin lock time
  • a smaller acceleration factor is used for variable-density acceleration sampling at a later time, and at a later larger TSL time point, a larger acceleration factor is used for variable-density acceleration during acquisition.
  • the sampling template in the frequency encoding direction is full sampling; the sampling template in the phase encoding direction is variable density sampling, and the sampling data in the phase encoding-parameter space conforms to the random sampling theory of compressed sensing.
  • the preset Fourier transform is used to convert the magnetic resonance data to the image domain to obtain an initial image, and the current compensation coefficient is determined based on the double exponential relaxation model and the initial image.
  • this embodiment converts the magnetic resonance data in the center part of K-space based on a preset Fourier transform to Image domain to get the initial image.
  • the influence of the coil sensitivity is not considered in the determination process of the initial image.
  • the initial image is nonlinearly fitted based on the double exponential relaxation model, the initial T 1 ⁇ s parameter and T 1 ⁇ l parameter are estimated, the initial compensation coefficient is obtained according to the T 1 ⁇ s parameter and T 1 ⁇ l parameter, and the initial image details are empty set.
  • the initial image is compensated based on the current compensation coefficient to obtain the compensated initial image, and then the image reconstruction is performed on the compensated initial image based on the L+S model, and the image generated in each iteration of the image reconstruction is named the intermediate image.
  • An intermediate image is divided by the current compensation coefficient to update the intermediate image, the T 1 ⁇ s parameter and T 1 ⁇ l parameter are updated based on the updated intermediate image, and then the current compensation coefficient is determined according to the updated T 1 ⁇ s parameter and T 1 ⁇ l parameter.
  • the loop process is: compensate the current intermediate image based on the current compensation coefficient to obtain the compensated current intermediate image.
  • the compensated current intermediate image is used as the input image of the next iteration, and the intermediate image generated by the next iteration is divided by the compensation coefficient to obtain the updated current intermediate image.
  • the above process is repeated until the iteration converges and the last iteration
  • the generated intermediate image is used as a parameter-weighted image.
  • the operations performed by the L+S model for each input image of the second iteration include: performing a singular value threshold operation and a soft threshold operation on the low-rank portion and the sparse portion of the compensated current intermediate image, respectively, to obtain the updated The low-rank part and sparse part of; then determine the current intermediate image according to the updated low-rank part and sparse part.
  • performing the soft threshold operation on the sparse portion of the compensated current intermediate image is: based on the current image details, performing a soft threshold operation on the sparse portion of the compensated current intermediate image to update the sparse portion. After the sparse part is updated, the image determined according to the updated sparse part and the updated low rank part is divided by the compensation coefficient to obtain the current intermediate image; the image details of the current intermediate image are extracted by the iterative detail operator to be used for compensation After the sparse part of the current intermediate image, a soft threshold operation is performed.
  • the singular value threshold operation on the low-rank part of the compensated current intermediate image is:
  • the double-exponential relaxation model is used to nonlinearly fit the parameter-weighted image to obtain the T 1 ⁇ s parameter and T 1 ⁇ l parameter, and then the T 1 ⁇ s parameter map and the T 1 ⁇ l parameter map.
  • the signal organization is usually composed of different parts of the interaction, and each part is composed of different protons, compared with the single exponential relaxation model that describes a single component, it is used to describe the double exponential relaxation of the two components.
  • the model can more accurately describe free water protons (such as extracellular water protons) and bound water protons (such as intracellular water protons) and their interactions.
  • the technical solution of the magnetic resonance parameter imaging method can acquire the magnetic resonance data of the target object in an under-acquisition manner, which can increase the scanning speed of the magnetic resonance data, and uses the preset Fourier transform to convert the magnetic resonance data to an image Field to get the initial image, determine the current compensation coefficient based on the double exponential relaxation model and the initial image; compensate the initial image based on the current compensation coefficient to obtain the compensated initial image, and input the compensated initial image as the input image for the first iteration L+S model, according to the L+S model in the image reconstruction process, iteratively generates the current intermediate image by updating the current compensation coefficient, and uses the updated current compensation coefficient to compensate the current intermediate image to generate the input compensation for the next iteration process After the current intermediate image is converged until the iteration, the intermediate image generated by the last iteration is used as the parameter-weighted image; the double-exponential relaxation model is used to nonlinearly fit the parameter-weighted image to obtain the parameter map.
  • the double exponential relaxation model can more accurately represent the change trend of tissues over time, especially for some complex tissues that have different proton components and interact with each other.
  • the exponential model is inaccurate to describe.
  • the T 1 ⁇ s parameter map and T 1 ⁇ l parameter map obtained by the double exponential relaxation model can better describe the interaction between free water protons and bound water protons, and it is more helpful to explore The underlying biophysical mechanism of T 1 ⁇ relaxation.
  • FIG. 1 is a flowchart of a magnetic resonance parameter imaging method provided by Embodiment 2 of the present invention.
  • the embodiment of the present invention is an optimization of the basis of the above embodiment. Accordingly, the method of this embodiment includes:
  • the preset Fourier transform is used to convert the magnetic resonance data to the image domain to obtain an initial image, and the current compensation coefficient is determined based on the double exponential relaxation model and the initial image.
  • the double exponential relaxation model is:
  • M represents the image intensity under different spin-lock time
  • M 0 represents the reference balanced image intensity without spin-lock pulse
  • ⁇ is the proportion of the long relaxation part
  • (1- ⁇ ) is The proportion of the short relaxation part
  • TSL k is the k-th spin-lock time
  • T 1 ⁇ s parameter and T 1 ⁇ l parameter are short T 1 ⁇ and long T 1 ⁇
  • N is the total number of spin-lock time Number
  • the T 1 ⁇ s parameter and T 1 ⁇ l parameter corresponding to each pixel can be obtained by nonlinearly fitting each pixel of the intermediate image through the above-mentioned double exponential relaxation model.
  • the nonlinear fitting here uses a trust region algorithm.
  • Coef is the compensation coefficient.
  • this embodiment Since the magnetic resonance data acquired in the under-recovery mode is full-recovery data in the center part of K-space, in order to quickly obtain a high-precision parameter map, this embodiment first converts the magnetic resonance data in the center part of K-space through a preset Fourier transform Go to the image field to get the initial image. Then, the initial T 1 ⁇ s parameter and T 1 ⁇ l parameter are obtained by nonlinearly fitting the initial image through the double exponential relaxation model, and then the current compensation coefficient for the coefficient compensation is obtained according to the coefficient compensation formula. This embodiment also initializes the image details to an empty set based on the initial image.
  • the L+S model is a low-rank plus sparse model (low-rank plus sparse, L+S), specifically:
  • ⁇ 1 is the l 1 norm
  • C( ⁇ ) is an operation operator, which means that pixel-level signal compensation is performed on the image
  • X is the image sequence to be reconstructed, and it is expressed as the size of the number of voxels ⁇ TSL (N) matrix
  • L is the low rank part expressed in matrix form
  • S is the sparse part expressed in matrix form, indicating the residual image between the image and the low rank part L
  • E is the multi-channel coil coding matrix, which It is equal to the product of the under-collected Fourier operator and the coil sensitivity matrix
  • Rank(L) is the rank of the low-rank part L
  • d is the K-space data obtained by under-collection.
  • the embodiment refers to the initial image and the image generated in each iteration as process images.
  • L j is the low-rank part of the current process image with the label j
  • S j is the sparse part of the current process image after the compensation with the label j.
  • Update Sj determine whether the image detail feat i is empty, if the image detail is extracted, then perform a soft threshold operation on the sparse part S according to the value in the image detail feat i , specifically: Among them, ST ( ⁇ ) is a soft threshold operation operator, defined as: Among them, P is an element of the current process image after compensation, and v is a threshold, the value of which is linearly related to the image detail feat i .
  • E * represents the inverse operation of E, that is, equal to the inverse Fourier transform of the K-space data of the multi-channel coil and then the coil combination to obtain the current process image
  • C -1 ( ⁇ ) means dividing each pixel of the image by the compensation coefficient.
  • IFR iterative feature refinement
  • the embodiment of the present invention uses the prior information of the double exponential relaxation model to increase the data redundancy of the parameter-weighted image, associates the independent reconstruction and fitting processes in the traditional fast parameter imaging method, and improves the accuracy of the parameter map .
  • the T 1 ⁇ s parameter map and T 1 ⁇ 1 parameter map corresponding to the double exponential relaxation model can better describe the interaction between free water protons and bound water protons. Function, it is more helpful to discover the underlying biophysical mechanism of T 1 ⁇ relaxation.
  • FIG. 2 is a structural block diagram of a magnetic resonance parameter imaging apparatus provided in Embodiment 4 of the present invention.
  • the device is used to execute the magnetic resonance parameter imaging method provided by any of the above embodiments, and the control device may be implemented by software or hardware.
  • the device includes:
  • the data acquisition module 11 is configured to acquire the magnetic resonance data of the target object in an under-acquisition manner
  • the initial image module 12 is set to convert the magnetic resonance data to an image domain using a preset Fourier transform to obtain an initial image, and determine a current compensation coefficient based on a double exponential relaxation model and the initial image;
  • the parameter-weighted image determination module 13 is set to compensate the initial image based on the current compensation coefficient to obtain the compensated initial image, and input the compensated initial image as the input image of the first iteration to the L+S model.
  • the current intermediate image is updated by iteratively generating the current compensation coefficient, and the updated current compensation coefficient is used to compensate the current intermediate image to generate the current intermediate image after the input of the next iteration process until the iteration convergence, and Use the intermediate image generated in the last iteration as the parameter-weighted image;
  • the parameter map determination module 14 is configured to use the double exponential relaxation model to perform a nonlinear fit on the parameter-weighted image to obtain a parameter map.
  • the technical solution of the magnetic resonance parameter imaging device provided by the embodiment of the present invention can use the prior information of the double exponential relaxation model to increase the data redundancy of the parameter-weighted image.
  • the independent reconstruction and fitting processes are linked, improving the accuracy of the parameter map.
  • the T 1 ⁇ s parameter map and T 1 ⁇ 1 parameter map corresponding to the double exponential relaxation model can better describe the interaction between free water protons and bound water protons. Function, it is more helpful to discover the underlying biophysical mechanism of T 1 ⁇ relaxation.
  • the magnetic resonance parameter imaging apparatus provided by the embodiment of the present invention can execute the magnetic resonance parameter imaging method provided by any embodiment of the present application, and has a function module and beneficial effects corresponding to the execution method.
  • FIG. 3 is a schematic structural diagram of a magnetic resonance device according to Embodiment 4 of the present invention.
  • the device includes a processor 201, a memory 202, an input device 203, and an output device 204; the number of processors 201 in the device may be One or more, one processor 201 is taken as an example in FIG. 3; the processor 201, the memory 202, the input device 203, and the output device 204 in the device may be connected by a bus or other means, and FIG. 3 takes the connection by a bus as an example .
  • the memory 202 can be used to store software programs, computer executable programs, and modules, such as program instructions/modules corresponding to the magnetic resonance parameter imaging method in the embodiments of the present invention (for example, the data acquisition module 11, Initial image module 12, parameter weighted image determination module 13, and parameter map determination module 14).
  • the processor 201 executes various functional applications and data processing of the device by running software programs, instructions, and modules stored in the memory 202, that is, implementing the above-mentioned magnetic resonance parameter imaging method.
  • the memory 202 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system and application programs required for at least one function; the storage data area may store data created according to the use of the terminal, and the like.
  • the memory 202 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.
  • the memory 202 may further include memories remotely provided with respect to the processor 201, and these remote memories may be connected to the device through a network. Examples of the above network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.
  • the input device 203 can be used to receive input digital or character information, and generate key signal input related to user settings and function control of the device.
  • the output device 204 may include a display device such as a display screen, for example, a display screen of a user terminal.
  • Embodiment 5 of the present invention further provides a storage medium containing computer-executable instructions, which when executed by a computer processor are used to perform a magnetic resonance parameter imaging method, the method including:
  • a preset Fourier transform is used to convert the magnetic resonance data to an image domain to obtain an initial image, and a current compensation coefficient is determined based on a double exponential relaxation model and the initial image;
  • the initial image is compensated based on the current compensation coefficient to obtain the compensated initial image.
  • the compensated initial image is input to the L+S model as the input image of the first iteration, and the current generated by iteration during the image reconstruction process according to the L+S model Update the current compensation coefficient of the intermediate image, and use the updated current compensation coefficient to compensate the current intermediate image to generate the current intermediate image after the input of the next iteration process until the iteration convergence, and the intermediate image generated by the last iteration Weighted image as a parameter;
  • the parameter-weighted image is nonlinearly fitted to obtain a parameter map.
  • a storage medium containing computer-executable instructions provided by an embodiment of the present invention is not limited to the method operations described above, and can also perform magnetic resonance parameter imaging provided by any embodiment of the present application. Related operations in the method.
  • the technical solution of the present application can essentially be embodied in the form of a software product that contributes to the related technology, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, Read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including several instructions to make a computer device (which can be a personal computer, Server, or network equipment, etc.) to execute the magnetic resonance parameter imaging methods described in the embodiments of the present application.
  • a computer-readable storage medium such as a computer floppy disk, Read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

一种磁共振参数成像方法、装置、设备及存储介质，该方法包括：将获取的磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和初始图像确定当前补偿系数；基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像输入L+S模型，根据L+S模型在迭代过程中生成的当前中间图像更新当前补偿系数，利用更新后的当前补偿系数对当前中间图像进行补偿以生成用于下一次迭代的补偿后的中间图像，迭代收敛，将最后一次迭代生成的中间图像作为参数加权图像；采用双指数弛豫模型对参数加权图像进行非线性拟合。

Description

磁共振参数成像方法、装置、设备及存储介质
本申请要求在2018年12月12日提交中国专利局、申请号为201811518375.1的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本申请涉及图像处理领域，例如涉及一种磁共振参数成像方法、装置、设备及存储介质。
背景技术
近年来，由于旋转坐标系下的纵向弛豫(spin-lattice relaxation in the rotating frame，T 1ρ)在骨关节炎中的软骨退变、膝盖软骨损伤、椎间盘退变、肝纤维化以及一些脑相关疾病诊断上的有效性，受到了临床上的广泛关注。在进行T 1ρ定量成像时，为了获得较高质量的T 1ρ参数图，通常一次扫描需要采集多幅不同自旋锁定时间(spin-lock time，TSL)的图像，因此其扫描时间往往很长。
为了加快扫描速度，减少扫描时间，相关技术要么减少TSL的数量，要么采用快速成像技术。由于T 1ρ参数图是由T 1ρ加权图像确定的，那么对于前者，由于TSL的减少，导致T 1ρ参数加权图像数量减少，进而导致T 1ρ参数图的精度降低，即导致定量精度的降低；对于后者，目前商用的快速成像技术主要是并行成像(如敏感度编码(SENSE)、广义自动校准部分并行采集(GRAPPA)等)，由于受并行成像列阵线圈的限制，加速倍数越高，其获得的T 1ρ参数加权图像的信噪比就越低，T 1ρ参数图的质量的精度就越低，即定量精度就越低，因此并行成像的扫描速度通常仅能达到2-3倍。
综上所述，相关技术在参数成像方法很难同时兼顾成像速度与成像质量。
发明内容
本发明实施例提供了一种磁共振参数成像方法、装置、设备及存储介质，以解决相关技术在参数成像方法存在很难同时兼顾成像速度与成像质量的技术问题。
本发明实施例提供了一种磁共振参数成像方法，包括：
以欠采方式获取目标对象的磁共振数据；
采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型(low-rank plus sparse，L+S，即低秩加稀疏模型)，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
本发明实施例还提供了一种磁共振参数成像装置，包括：
数据获取模块，设置为以欠采方式获取目标对象的磁共振数据；
初始图像模块，设置为采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
参数加权图像确定模块，设置为基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前 补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
参数图确定模块，设置为采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
本发明实施例还提供了一种磁共振设备，包括：
一个或多个处理器；
存储装置，用于存储一个或多个程序；
当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如上所述的磁共振参数成像方法。
本发明实施例还提供了一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行如上所述的磁共振参数成像方法。
本发明实施例提供的磁共振参数成像方法的技术方案，以欠采方式获取目标对象的磁共振数据，可以提高磁共振数据扫描的速度，采用预设傅里叶变换将磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和初始图像确定当前补偿系数；基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；采用双指数弛豫模型对参数加权图像进行非线性拟合以得到参数图。相较于单指数弛豫模型，双指数弛豫模型能更为准确地表示组织 随时间的变化趋势，尤其是对于存在着不同的质子组成部分且组成部分相互作用的一些复杂的组织，用单指数模型来描述是不准确的，通过双指数弛豫模型拟合得到的T 1ρs参数图和T 1ρl参数图可以更好地描述自由水质子和受束缚水质子的相互作用，更有助于发掘T 1ρ弛豫的底层生物物理机制。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图做一简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本发明实施例一提供的磁共振参数成像方法的流程图；
图2是本发明实施例三提供的磁共振参数成像装置结构框图；
图3是本发明实施例四提供的磁共振设备的结构框图。
具体实施方式
以下将参照本发明实施例中的附图，通过实施方式清楚、完整地描述本申请的技术方案，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
实施例一
图1是本发明实施例一提供的磁共振参数成像方法的流程图。本实施例的技术方案适用于快速获取高质量的参数图的情况。该方法可以由本发明实施例提供的磁共振参数成像装置来执行，该装置可以采用软件和/或硬件的方式实现， 并配置在处理器中应用。该方法具体包括如下步骤：
S101、以欠采方式获取目标对象的磁共振数据。
为了提高磁共振参数图的成像速度，本实施例基于稀疏采样理论，以欠采方式获取目标对象的磁共振数据，具体为：由于参数图像往往需要采集多个TSL(spin lock time)时间点的T 1ρ参数加权图像，对于小的TSL时间点，采集时采用较小的加速倍数进行变密度加速采样，在稍后的较大的TSL时间点，采集时采用较大的加速倍数进行变密度加速采样，从而获得欠采的K空间数据。另外，频率编码方向的采样模板为全采样；相位编码方向的采样模板为变密度采样，且相位编码-参数空间的采样数据符合压缩感知的随机采样理论。
S102、采用预设傅里叶变换将磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和初始图像确定当前补偿系数。
由于以欠采方式获取的磁共振数据在K空间中心部分为全采数据，为了快速得到高精度的参数图，本实施例基于预设傅里叶变换将K空间中心部分的磁共振数据转换至图像域以得到初始图像。本实施例在初始图像的确定过程不考虑线圈敏感度的影响。
初始图像确定后，基于双指数弛豫模型对初始图像进行非线性拟合，估计初始的T 1ρs参数和T 1ρl参数，根据T 1ρs参数和T 1ρl参数得到初始补偿系数，并初始化图像细节为空集。
S103、基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为 参数加权图像。
基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，然后基于L+S模型对补偿后的初始图像进行图像重建，将图像重建中每次迭代生成的图像命名为中间图像，将每个中间图像除以当前补偿系数以更新中间图像，基于更新后的中间图像更新T 1ρs参数和T 1ρl参数，然后根据更新后的T 1ρs参数和T 1ρl参数确定当前补偿系数。一些实施例中，该循环过程为：基于当前补偿系数补偿当前中间图像以得到补偿后的当前中间图像。将补偿后的当前中间图像作为下一次迭代的输入图像，同时使用该下一次迭代生成的中间图像除以补偿系数得到更新后的当前中间图像，重复上述过程，直至迭代收敛，将最后一次迭代所生成的中间图像作为参数加权图像。
一些实施例中，L+S模型每对次迭代的输入图像执行的操作包括：对补偿后的当前中间图像的低秩部分和稀疏部分别做奇异值阈值操作和软阈值操作，以得到更新后的低秩部分和稀疏部分；然后根据更新后的低秩部分和稀疏部分确定当前中间图像。
一些实施例中，对补偿后的当前中间图像的稀疏部分做软阈值操作为：基于当前的图像细节，对补偿后的当前中间图像的稀疏部分进行软阈值操作以更新稀疏部分。稀疏部分更新后，将根据更新后的稀疏部分和更新后的低秩部分所确定的图像除以补偿系数得到当前中间图像；通过迭代细节算子提取当前中间图像的图像细节，以用于对补偿后的当前中间图像的稀疏部分做软阈值操作。
一些实施例中，对补偿后的当前中间图像的低秩部分做奇异值阈值操作为：
Figure PCTCN2019123448-appb-000001
其中，L j为标号为j的补偿后的当前中间图像的低秩部分，SVT(·)表示表示奇异值阈值操作算子，定义为：SVT λ(M)＝UΛ λ(Σ)V H，其中，M＝UΣV H表示奇 异值分解(SVD)，U和V分别为左、右奇异值向量组成的矩阵，V H表示V的共轭转置，Σ是由M的奇异值组成的对角矩阵，Λ λ(Σ)表示保留Σ中最大的奇异值不变，其他全为0，S j为标号为j的补偿后的当前中间图像的稀疏部分。
S104、采用双指数弛豫模型对参数加权图像进行非线性拟合以得到参数图。
得到参数加权图像后，采用双指数弛豫模型对参数加权图像进行非线性拟合，得到T 1ρs参数和T 1ρl参数，进而得到T 1ρs参数图和T 1ρl参数图。由于信号组织通常是由相互作用的不同部分组成，且每个部分由不同的质子组成，因此，相对于描述单一成分的单指数弛豫模型来说，用于描述两种成分的双指数弛豫模型更能准确地描述自由水质子(如细胞外的水质子)和受束缚水质子(如细胞内的水质子)以及它们之间的相互作用。
本发明实施例提供的磁共振参数成像方法的技术方案，以欠采方式获取目标对象的磁共振数据，可以提高磁共振数据扫描的速度，采用预设傅里叶变换将磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和初始图像确定当前补偿系数；基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；采用双指数弛豫模型对参数加权图像进行非线性拟合以得到参数图。相较于单指数弛豫模型，双指数弛豫模型能更为准确地表示组织随时间的变化趋势，尤其是对于存在着不同的质子组成部分且组成部分相互作用的一些复杂的组织，用单指数模型来描述是不准确的，通过双指数弛豫模型拟合得到的T 1ρs参数图和T 1ρl参数图可以更好地描述自由水质子和受束缚水质子 的相互作用，更有助于发掘T 1ρ弛豫的底层生物物理机制。
实施例二
图1是本发明实施例二提供的磁共振参数成像方法的流程图。本发明实施例是对上述实施例的基础的优化。相应地，本实施例的方法包括：
S101、以欠采方式获取目标对象的磁共振数据。
S102、采用预设傅里叶变换将磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和初始图像确定当前补偿系数。
双指数弛豫模型为：
Figure PCTCN2019123448-appb-000002
其中，M表示不同自旋-锁时间下的图像强度；M 0表示不带自旋-锁脉冲情况下的基准平衡图像强度，α是长弛豫部分所占的比例，(1-α)为短弛豫部分所占的比例，TSL k是第k个自旋-锁时间，T 1ρs参数和T 1ρl参数分别为短的T 1ρ和长的T 1ρ，N是自旋-锁定时间的总个数；0.1＜α＜1，T 1ρs≤40ms，40ms＜T 1ρl≤200ms。通过上述双指数弛豫模型对中间图像的每个像素进行非线性拟合可以得到每个像素对应的T 1ρs参数和T 1ρl参数，此处的非线性拟合采用了信赖域算法。
由于信号补偿可表示为将图像中的每个像素乘以一个补偿系数，因此根据前述双指数弛豫模型可得知系数补偿公式为：
Figure PCTCN2019123448-appb-000003
其中，Coef为补偿系数。
由于以欠采方式获取的磁共振数据在K空间中心部分为全采数据，为了快速得到高精度的参数图，本实施例先将K空间中心部分的磁共振数据通过预设 傅里叶变换转换至图像域以得到初始图像。然后通过双指数弛豫模型对初始图像进行非线性拟合得到初始的T 1ρs参数和T 1ρl参数，然后根据系数补偿公式得到用于系数补偿的当前补偿系数。本实施例还根据初始图像将图像细节初始化为空集。
S103、基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像。
其中，L+S模型为低秩加稀疏模型(low-rank plus sparse，L+S)，具体为：
min {X,L,S}‖S‖ 1 S.T.C(X)＝L+S,E(X)＝d,Rank(L)＝1
其中，‖·‖ 1是l 1范数；C(·)是一个操作算子，表示对图像进行像素级信号补偿，X为待重建的图像序列，且其表示成大小为体素数×TSL数(N)的矩阵；L是用矩阵形式表示的低秩部分；S是用矩阵形式表示的稀疏部分，表示图像与低秩部分L之间的残差图像；E是多通道线圈编码矩阵，其等于欠采傅里叶算子与线圈敏感度矩阵的乘积；Rank(L)为低秩部分L的秩，d是以欠采方式获取的K空间数据。
由于初始图像和L+S模型的系数补偿方法相同，以及L+S模型对初始图像和L+S模型迭代生成的中间图像所进行的图像处理方法相同，因此，为了便于技术方案的阐述，本实施例将初始图像和每次迭代生成的图像均称为过程图像。
设定循环i＝1，2…，在i次的迭代中：
1)、根据当前补偿系数对当前过程图像进行补偿以得到补偿后的当前过程 图像，具体为：
Figure PCTCN2019123448-appb-000004
其中，U表示补偿后的当前过程图像，如果i为1，则当前过程图像为初始图像，如果i为1，则当前过程图像为中间图像。
2)、确定补偿后的当前过程图像的低秩部分L和稀疏部分S，初始化S＝0，设定内循环次数为j，在第j＝1，2，…,在j次的迭代中：
a)、更新L j：
Figure PCTCN2019123448-appb-000005
其中，L j为标号为j的补偿后的当前过程图像的低秩部分，SVT(·)表示奇异值阈值操作算子，定义为：SVT λ(M)＝UΛ λ(Σ)V H，其中，M＝UΣV H表示奇异值分解(SVD)，U和V分别为左、右奇异值向量组成的矩阵，V H表示V的共轭转置，Σ是由M的奇异值组成的对角矩阵，Λ λ(Σ)表示保留Σ中最大的奇异值不变，其他全为0，本实施例仅取L的最大奇异值，使得奇异值阈值操作后L的秩Rank(L)＝1；S j为标号为j的补偿后的当前过程图像的稀疏部分。
b)、更新Sj：判断图像细节feat i是否为空，如果有提取到图像细节，则根据图像细节feat i中的值来对稀疏部分S进行软阈值操作，具体为：
Figure PCTCN2019123448-appb-000006
其中，ST(·)是软阈值操作算子，定义为：
Figure PCTCN2019123448-appb-000007
其中，P是补偿后的当前过程图像的一个元素，v是阈值，其值与图像细节feat i线性相关。通过对稀疏矩阵S做软阈值操作，可以有效地去除图像伪影。
c)、更新数据保真项：
Figure PCTCN2019123448-appb-000008
其中E *表示E的逆操作，即等于对多通道线圈的K空间数据做傅里叶逆变换后再进行线圈组合以得到当前过程图像；
d)、更新当前过程图像X i，
Figure PCTCN2019123448-appb-000009
其中，C -1(·)表示将图像的每个像素除以补偿系数。
e)、终止内循环迭代。
3)、根据更新后的当前过程图像，结合双指数弛豫模型更新T 1ρs参数和T 1ρ1参数，根据更新后的T 1ρs参数和T 1ρ1参数更新补偿系数，以用于补偿更新后的当前过程图像；以及利用迭代细节提取算子(iterative feature refinement，IFR)提取更新后的当前过程图像的图像细节，以用于在新一次迭代过冲中对补偿后的当前过程图像的稀疏矩阵做软阈值操作。
4)、算法收敛，终止循环迭代，以得到最终的T 1ρ参数加权图像。
S104、采用双指数弛豫模型对参数加权图像进行非线性拟合以得到参数图。
本发明实施例利用双指数弛豫模型的先验信息来增加参数加权图像的数据冗余性，将传统快速参数成像方法中相互独立的重建和拟合过程关联起来，提高了参数图的精确度。而且相较于相关技术中单指数弛豫模型对应的T 1ρ参数图，双指数弛豫模型对应的T 1ρs参数图和T 1ρ1参数图可以更好地描述自由水质子和受束缚水质子的相互作用，更有助于发掘T 1ρ弛豫的底层生物物理机制。
实施例三
图2是本发明实施例四提供的磁共振参数成像装置的结构框图。该装置用于执行上述任意实施例所提供的磁共振参数成像方法，该控制装置可选为软件或硬件实现。该装置包括：
数据获取模块11，设置为以欠采方式获取目标对象的磁共振数据；
初始图像模块12，设置为采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
参数加权图像确定模块13，设置为基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入 L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
参数图确定模块14，设置为采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
本发明实施例提供的磁共振参数成像装置的技术方案，相较于相关技术，可以利用双指数弛豫模型的先验信息来增加参数加权图像的数据冗余性，将传统快速参数成像方法中相互独立的重建和拟合过程关联起来，提高了参数图的精确度。而且相较于相关技术中单指数弛豫模型对应的T 1ρ参数图，双指数弛豫模型对应的T 1ρs参数图和T 1ρ1参数图可以更好地描述自由水质子和受束缚水质子的相互作用，更有助于发掘T 1ρ弛豫的底层生物物理机制。
本发明实施例所提供的磁共振参数成像装置可执行本申请任一实施例所提供的磁共振参数成像方法，具备执行方法相应的功能模块和有益效果。
实施例四
图3为本发明实施例四提供的磁共振设备的结构示意图，如图3所示，该设备包括处理器201、存储器202、输入装置203以及输出装置204；设备中处理器201的数量可以是一个或多个，图3中以一个处理器201为例；设备中的处理器201、存储器202、输入装置203以及输出装置204可以通过总线或其他方式连接，图3中以通过总线连接为例。
存储器202作为一种计算机可读存储介质，可用于存储软件程序、计算机可执行程序以及模块，如本发明实施例中的磁共振参数成像方法对应的程序指 令/模块(例如，数据获取模块11、初始图像模块12、参数加权图像确定模块13以及参数图确定模块14)。处理器201通过运行存储在存储器202中的软件程序、指令以及模块，从而执行设备的各种功能应用以及数据处理，即实现上述的磁共振参数成像方法。
存储器202可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序；存储数据区可存储根据终端的使用所创建的数据等。此外，存储器202可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。在一些实例中，存储器202可进一步包括相对于处理器201远程设置的存储器，这些远程存储器可以通过网络连接至设备。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入装置203可用于接收输入的数字或字符信息，以及产生与设备的用户设置以及功能控制有关的键信号输入。
输出装置204可包括显示屏等显示设备，例如，用户终端的显示屏。
实施例五
本发明实施例五还提供一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行磁共振参数成像方法，该方法包括：
以欠采方式获取目标对象的磁共振数据；
采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿 后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
当然，本发明实施例所提供的一种包含计算机可执行指令的存储介质，其计算机可执行指令不限于如上所述的方法操作，还可以执行本申请任一实施例所提供的磁共振参数成像方法中的相关操作。
通过以上关于实施方式的描述，所属领域的技术人员可以清楚地了解到，本申请可借助软件及必需的通用硬件来实现，当然也可以通过硬件实现，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对相关技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中，如计算机的软盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、闪存(FLASH)、硬盘或光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述的磁共振参数成像方法。
值得注意的是，上述磁共振参数成像装置的实施例中，所包括的各个单元和模块只是按照功能逻辑进行划分的，但并不局限于上述的划分，只要能够实现相应的功能即可；另外，各功能单元的具体名称也只是为了便于相互区分，并不用于限制本申请的保护范围。
注意，上述仅为本申请的可选实施例及所运用技术原理。本领域技术人员会理解，本申请不限于这里所述的可选实施例，对本领域技术人员来说能够进行各种明显的变化、重新调整和替代而不会脱离本申请的保护范围。因此，虽然通过以上实施例对本申请进行了较为详细的说明，但是本申请不仅仅限于以上实施例，在不脱离本申请构思的情况下，还可以包括更多其他等效实施例，而本申请的范围由所附的权利要求范围决定。

Claims (10)

  1. 一种磁共振参数成像方法，包括：
    以欠采方式获取目标对象的磁共振数据；
    采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
    基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
    采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
  2. 根据权利要求1所述的方法，其中，所述以欠采方式获取目标对象的磁共振数据，包括：
    基于稀疏采样理论，以变速率变密度的欠采方式获取目标对象的磁共振数据。
  3. 根据权利要求2所述的方法，其中，变速率变密度的欠采方式为：
    欠采倍数随着时间点的递增而递增；
    频率编码方向的采样模板为全采样；
    相位编码方向的采样模板为变密度采样，且相位编码-参数空间的采样数据符合压缩感知的随机采样理论。
  4. 根据权利要求1-3任一项所述的方法，其中，所述根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的 当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像，包括：
    基于双指数弛豫模型和当前中间图像确定当前补偿系数；
    使用当前补偿系数对当前中间图像进行补偿以得到补偿后的当前中间图像；
    将补偿后的当前中间图像输入新一次迭代，以对补偿后的当前中间图像的低秩部分和稀疏部分别做奇异值阈值操作和软阈值操作，以更新低秩部分和稀疏部分；
    将更新后的低秩部分和稀疏部分所确定的图像除以补偿系数以得到当前中间图像；
    重复上述过程，直至迭代收敛，并将最后一次迭代生成的中间图像作为参数加权图像。
  5. 根据权利要求4所述的方法，其中，对补偿后的中间图像的稀疏部分做软阈值操作，包括：
    根据当前中间图像的图像细节对补偿后的当前中间图像的稀疏部分进行软阈值操作以更新所述稀疏部分。
  6. 根据权利要求4所述的方法，其特征在于，对补偿后的中间图像的低秩部分做奇异值阈值操作为：
    Figure PCTCN2019123448-appb-100001
    其中，L j为标号为j的补偿后的当前中间图像的低秩部分，SVT(·)表示奇异值阈值操作算子，定义为：SVT λ(M)＝UΛ λ(Σ)V H，其中，M＝UΣV H表示奇异值分解(SVD)，U和V分别为左、右奇异值向量组成的矩阵，V H表示V的共轭 转置，Σ是由M的奇异值组成的对角矩阵，Λ λ(Σ)表示保留Σ中最大的奇异值不变，其他全为0，S j为标号为j的补偿后的当前中间图像的稀疏部分。
  7. 根据权利要求6述的方法，其中，所述双指数弛豫模型为：
    Figure PCTCN2019123448-appb-100002
    其中，M表示不同自旋-锁时间下的图像强度；M 0表示不带自旋-锁脉冲情况下的基准平衡图像强度，α是长弛豫部分所占的比例，(1-α)为短弛豫部分所占的比例，且0.1＜α＜1，TSL k是第k个自旋-锁时间，T 1ρs和T 1ρl分别为短的T 1ρ和长的T 1ρ，且T 1ρs≤40ms，40ms＜T 1ρl≤200ms，N是自旋-锁定时间的总个数。
  8. 一种磁共振参数成像装置，包括：
    数据获取模块，设置为以欠采方式获取目标对象的磁共振数据；
    初始图像模块，设置为采用预设傅里叶变换将所述磁共振数据转换至图像域以得到初始图像，基于双指数弛豫模型和所述初始图像确定当前补偿系数；
    参数加权图像确定模块，设置为基于当前补偿系数对初始图像进行补偿以得到补偿后的初始图像，将补偿后的初始图像作为首次迭代的输入图像输入L+S模型，根据L+S模型在图像重建过程中通过迭代生成的当前中间图像更新当前补偿系数，并利用更新后的当前补偿系数对当前中间图像进行补偿以生成输入下一次迭代过程的补偿后的当前中间图像，直至迭代收敛，并将最后一次迭代所生成的中间图像作为参数加权图像；
    参数图确定模块，设置为采用所述双指数弛豫模型对所述参数加权图像进行非线性拟合以得到参数图。
  9. 一种磁共振设备，包括：
    一个或多个处理器；
    存储装置，用于存储一个或多个程序；
    当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如权利要求1-7中任一项所述的磁共振参数成像方法。
  10. 一种包含计算机可执行指令的存储介质，其中，所述计算机可执行指令在由计算机处理器执行时用于执行如权利要求1-7中任一项所述的磁共振参数成像方法。
PCT/CN2019/123448 2018-12-12 2019-12-05 磁共振参数成像方法、装置、设备及存储介质 WO2020119581A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811518375.1A CN109658468B (zh) 2018-12-12 2018-12-12 磁共振参数成像方法、装置、设备及存储介质
CN201811518375.1 2018-12-12

Publications (1)

Publication Number Publication Date
WO2020119581A1 true WO2020119581A1 (zh) 2020-06-18

Family

ID=66113811

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/123448 WO2020119581A1 (zh) 2018-12-12 2019-12-05 磁共振参数成像方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN109658468B (zh)
WO (1) WO2020119581A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112598130A (zh) * 2020-12-09 2021-04-02 华东交通大学 基于自编码器和奇异值阈值的土壤湿度数据重构方法和计算机可读存储介质

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109658468B (zh) * 2018-12-12 2022-12-13 深圳先进技术研究院 磁共振参数成像方法、装置、设备及存储介质
CN110889897B (zh) * 2019-11-21 2021-04-06 厦门大学 一种体素内不相干运动磁共振成像参数重建方法及系统
CN113050009B (zh) * 2019-12-27 2022-12-13 深圳先进技术研究院 三维磁共振快速参数成像方法和装置
WO2021217509A1 (zh) * 2020-04-29 2021-11-04 深圳先进技术研究院 一种磁共振旋转坐标系下的自旋晶格弛豫成像方法和系统
CN113567901A (zh) * 2020-04-29 2021-10-29 深圳先进技术研究院 一种磁共振旋转坐标系下的自旋晶格弛豫成像方法和系统
CN113052937B (zh) * 2021-04-19 2022-07-29 厦门大学 一种动态对比增强核磁共振图像的重建方法及重建系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102389309A (zh) * 2011-07-08 2012-03-28 首都医科大学 基于压缩感知理论的磁共振图像重建的方法
CN103400402A (zh) * 2013-07-12 2013-11-20 西安电子科技大学 基于低秩结构稀疏的压缩感知mri图像重建方法
CN103654789A (zh) * 2013-12-10 2014-03-26 深圳先进技术研究院 磁共振快速参数成像方法和系统
CN104871026A (zh) * 2012-12-17 2015-08-26 通用电气公司 靶物质的体外磁共振检测
CN109658468A (zh) * 2018-12-12 2019-04-19 深圳先进技术研究院 磁共振参数成像方法、装置、设备及存储介质

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103035017A (zh) * 2011-12-08 2013-04-10 中国科学院深圳先进技术研究院 磁共振参数成像方法和系统
US10393839B2 (en) * 2012-02-03 2019-08-27 Arizona Board Of Regents On Behalf Of The University Of Arizona System and method for image processing with highly undersampled imaging data
CN103472419B (zh) * 2013-08-30 2016-05-18 深圳先进技术研究院 磁共振快速成像方法及其系统
US10690740B2 (en) * 2014-11-01 2020-06-23 Mayo Foundation For Medical Education And Research Sparse reconstruction strategy for multi-level sampled MRI
CN108828483A (zh) * 2018-04-25 2018-11-16 东北大学 一种磁共振图像的收缩阈值迭代重建算法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102389309A (zh) * 2011-07-08 2012-03-28 首都医科大学 基于压缩感知理论的磁共振图像重建的方法
CN104871026A (zh) * 2012-12-17 2015-08-26 通用电气公司 靶物质的体外磁共振检测
CN103400402A (zh) * 2013-07-12 2013-11-20 西安电子科技大学 基于低秩结构稀疏的压缩感知mri图像重建方法
CN103654789A (zh) * 2013-12-10 2014-03-26 深圳先进技术研究院 磁共振快速参数成像方法和系统
CN109658468A (zh) * 2018-12-12 2019-04-19 深圳先进技术研究院 磁共振参数成像方法、装置、设备及存储介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112598130A (zh) * 2020-12-09 2021-04-02 华东交通大学 基于自编码器和奇异值阈值的土壤湿度数据重构方法和计算机可读存储介质
CN112598130B (zh) * 2020-12-09 2024-04-09 华东交通大学 基于自编码器和奇异值阈值的土壤湿度数据重构方法和计算机可读存储介质

Also Published As

Publication number Publication date
CN109658468B (zh) 2022-12-13
CN109658468A (zh) 2019-04-19

Similar Documents

Publication Publication Date Title
WO2020119581A1 (zh) 磁共振参数成像方法、装置、设备及存储介质
Güngör et al. Adaptive diffusion priors for accelerated MRI reconstruction
US10671939B2 (en) System, method and computer-accessible medium for learning an optimized variational network for medical image reconstruction
Liu et al. Projected iterative soft-thresholding algorithm for tight frames in compressed sensing magnetic resonance imaging
Fabian et al. Data augmentation for deep learning based accelerated MRI reconstruction with limited data
US11341616B2 (en) Methods and system for selective removal of streak artifacts and noise from images using deep neural networks
WO2020114329A1 (zh) 磁共振快速参数成像方法及装置
US8879852B2 (en) Non-contrast-enhanced 4D MRA using compressed sensing reconstruction
US11830111B2 (en) Magnetic resonance imaging method, device, medical device and storage medium
EP3346443B1 (en) Method for restoring magnetic resonance image and magnetic resonance image processing apparatus
US10180476B2 (en) Systems and methods for segmented magnetic resonance fingerprinting dictionary matching
Pal et al. A review and experimental evaluation of deep learning methods for MRI reconstruction
US20230097417A1 (en) Multi-slice magnetic resonance imaging method and device based on long-distance attention model reconstruction
Kelkar et al. Prior image-constrained reconstruction using style-based generative models
KR102428725B1 (ko) 영상 개선 방법 및 이를 수행하는 컴퓨터 프로그램
WO2021253671A1 (zh) 磁共振电影成像方法、装置、成像设备及存储介质
WO2024021796A1 (zh) 一种图像处理方法、装置、电子设备、存储介质及程序产品
Kamesh Iyer et al. Split Bregman multicoil accelerated reconstruction technique: A new framework for rapid reconstruction of cardiac perfusion MRI
KR102526487B1 (ko) 딥러닝 모델의 학습을 위한 의료 데이터의 처리 방법, 프로그램 및 장치
WO2021129235A1 (zh) 三维磁共振快速参数成像方法和装置
WO2012061475A2 (en) Systems and methods for fast magnetic resonance image reconstruction
Angshul A sparsity based approach towards fast MRI acquisition
Lin et al. Accelerated methods for low-rank plus sparse image reconstruction
CN113567901A (zh) 一种磁共振旋转坐标系下的自旋晶格弛豫成像方法和系统
US11965947B2 (en) Multiplex MRI image reconstruction

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19895196

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 03.11.2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19895196

Country of ref document: EP

Kind code of ref document: A1