WO2023115339A1 - 基于物理的数据增强对比学习表示成像方法 - Google Patents

基于物理的数据增强对比学习表示成像方法 Download PDF

Info

Publication number
WO2023115339A1
WO2023115339A1 PCT/CN2021/140119 CN2021140119W WO2023115339A1 WO 2023115339 A1 WO2023115339 A1 WO 2023115339A1 CN 2021140119 W CN2021140119 W CN 2021140119W WO 2023115339 A1 WO2023115339 A1 WO 2023115339A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
subset
contrastive
image
magnetic resonance
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2021/140119
Other languages
English (en)
French (fr)
Inventor
王珊珊
郑海荣
吴若有
刘新
梁栋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to PCT/CN2021/140119 priority Critical patent/WO2023115339A1/zh
Publication of WO2023115339A1 publication Critical patent/WO2023115339A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
    • A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—Two-dimensional [2D] image generation

Definitions

  • the invention relates to medical imaging technology, in particular to a physics-based data enhancement contrast learning representation imaging method, device, equipment and storage medium thereof.
  • Magnetic resonance imaging is widely used in the diagnosis of various diseases. Compared with other medical imaging techniques such as computed tomography (CT), one bottleneck of MRI is the long scanning time.
  • CT computed tomography
  • Parallel MRI is a widely used acceleration strategy that utilizes redundant information provided by multiple receive coils to reduce scan time.
  • the embodiment of the present application provides a physical-based data-enhanced contrastive learning representation imaging method, the method comprising: performing multi-physics under-sampled data enhancement on magnetic resonance data; and performing enhanced under-sampled magnetic resonance data
  • Use contrastive representation learning framework use contrastive loss constraints and data fitting items to ensure the accuracy of MRI.
  • the undersampling data enhancement of multi-physics trajectories for the magnetic resonance data includes: performing twice undersampling on the undersampled data to obtain the data enhancement subset 1 and the data enhancement subset 2, and the data enhancement Subset One and Data Augmentation Subset Two combine the sensitivity maps of the images to generate the input images for the network.
  • the contrastive representation learning framework is used for the enhanced under-sampled magnetic resonance data, including: constructing corresponding parallel network framework reconstruction network 1 and reconstruction network based on data enhancement subset 1 and data enhancement subset 2 2. Input the data augmentation subset 1 into the reconstruction network to generate the reconstruction image 1, and input the data augmentation subset 2 into the reconstruction network 2 to generate the reconstruction image 2.
  • the use of contrast loss constraints and data fitting items to ensure the accuracy of magnetic resonance imaging includes: performing reconstruction loss design for reconstructed image 1 and reconstructed image 2; comparing reconstructed image 1 and reconstructed image 2 The loss between the reconstructed image 1 and the reconstructed image 2 is guaranteed to be consistent; the data consistency loss design is performed on the transformed image 1 generated after the reconstructed image 1 and the transformed image 2 generated after the reconstructed image 2; the trained model For testing, the average of the output images of the parallel network is used as the final reconstruction result.
  • the embodiment of the present application also provides a physics-based data enhancement contrastive learning representation imaging device, which includes: an enhancement unit for performing multi-physics undersampling data enhancement on magnetic resonance data; a reconstruction unit, It is used to use the contrastive representation learning framework for the enhanced under-sampled magnetic resonance data; the guarantee unit is used to use contrastive loss constraints and data fitting items to ensure the accuracy of magnetic resonance imaging.
  • the undersampling data enhancement of multi-physics trajectories for the magnetic resonance data includes: performing twice undersampling on the undersampled data to obtain the data enhancement subset 1 and the data enhancement subset 2, and the data enhancement Subset One and Data Augmentation Subset Two combine the sensitivity maps of the images to generate the input images for the network.
  • the contrastive representation learning framework is used for the enhanced under-sampled magnetic resonance data, including: constructing corresponding parallel network framework reconstruction network 1 and reconstruction network based on data enhancement subset 1 and data enhancement subset 2 2. Input the data augmentation subset 1 into the reconstruction network to generate the reconstruction image 1, and input the data augmentation subset 2 into the reconstruction network 2 to generate the reconstruction image 2.
  • the use of contrast loss constraints and data fitting items to ensure the accuracy of magnetic resonance imaging includes: a reconstruction loss unit, which is used to design reconstruction loss for reconstructed image 1 and reconstructed image 2; contrast loss The unit is used to compare the loss between the reconstructed image 1 and the reconstructed image 2 to ensure that the outputs of the reconstructed image 1 and the reconstructed image 2 are consistent; the data consistency loss unit is used to reconstruct the transformed image 1 and reconstructed image 1 generated after the reconstructed image 1 The transformed image 2 generated after image 2 is designed for data consistency loss; the model test unit is used to test the trained model, and the average value of the output images of the parallel network is used as the final reconstruction result.
  • a reconstruction loss unit which is used to design reconstruction loss for reconstructed image 1 and reconstructed image 2
  • contrast loss The unit is used to compare the loss between the reconstructed image 1 and the reconstructed image 2 to ensure that the outputs of the reconstructed image 1 and the reconstructed image 2 are consistent
  • the data consistency loss unit is used to reconstruct the transformed
  • the embodiment of the present application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and operable on the processor.
  • the processor executes the program, it implements the The method described in any one of the descriptions of the examples.
  • the embodiment of the present application also provides a computer device, a computer-readable storage medium, on which a computer program is stored, and the computer program is used for: when the computer program is executed by a processor, the computer program according to the present application is implemented.
  • a computer device a computer-readable storage medium, on which a computer program is stored, and the computer program is used for: when the computer program is executed by a processor, the computer program according to the present application is implemented.
  • the physical-based data-enhanced contrastive learning representation imaging method provided by the invention solves the problem of dependence on full-sampled data and improves the utilization efficiency of under-sampled data.
  • the present invention proposes a parallel network framework for fast magnetic resonance imaging, which provides guidance for magnetic resonance imaging technology.
  • FIG. 1 shows a schematic flow diagram of a physics-based data-enhanced contrastive learning representation imaging method provided by an embodiment of the present application
  • FIG. 2 shows a schematic flowchart of a physics-based data-enhanced contrastive learning representation imaging method provided by another embodiment of the present application
  • FIG. 3 shows an exemplary structural block diagram of a physics-based data-enhanced contrastive learning representation imaging device 300 according to an embodiment of the present application
  • FIG. 4 shows an exemplary structural block diagram of a physics-based data-enhanced contrastive learning representation imaging device 400 according to another embodiment of the present application
  • FIG. 5 shows a schematic structural diagram of a computer system suitable for implementing a terminal device according to an embodiment of the present application
  • Fig. 6 shows another flowchart provided by the embodiment of the present application.
  • first and second are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features.
  • the features defined as “first” and “second” may explicitly or implicitly include at least one of these features.
  • “plurality” means at least two, such as two, three, etc., unless otherwise specifically defined.
  • the first feature may be in direct contact with the first feature or the first and second feature may be in direct contact with the second feature through an intermediary. touch.
  • “above”, “above” and “above” the first feature on the second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature.
  • “Below”, “beneath” and “beneath” the first feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is less horizontally than the second feature.
  • FIG. 1 shows a schematic flow chart of a physics-based data-enhanced contrastive learning representation imaging method provided by an embodiment of the present application.
  • the method includes:
  • Step 110 performing multi-physics undersampling data enhancement on the magnetic resonance data
  • Step 120 using a contrastive representation learning framework for the enhanced undersampled magnetic resonance data
  • Step 130 using contrast loss constraints and data fitting items to ensure the accuracy of magnetic resonance imaging.
  • the present invention proposes a parallel network framework for fast magnetic resonance imaging, which provides guidance for magnetic resonance imaging technology.
  • the multi-physics undersampling data enhancement of the magnetic resonance data in the present application includes: performing secondary undersampling on the undersampling data to obtain data enhancement subset 1 and data enhancement Subset two, combine data augmentation subset one and data augmentation subset two with the sensitivity map of the image to generate the input image of the network.
  • the contrastive representation learning framework for enhanced undersampled magnetic resonance data in the present application includes: building a corresponding parallel network based on data enhancement subset 1 and data enhancement subset 2 Frame reconstruction network 1 and reconstruction network 2; input data augmentation subset 1 into reconstruction network to generate reconstruction image 1, and input data augmentation subset 2 into reconstruction network 2 to generate reconstruction image 2.
  • FIG. 2 shows a schematic flowchart of a physics-based data-enhanced contrastive learning representation imaging method provided in another embodiment of the present application.
  • the contrast loss constraints and data fitting items are used to ensure the accuracy of magnetic resonance imaging, including:
  • Step 210 perform reconstruction loss design for reconstructed image 1 and reconstructed image 2;
  • Step 220 comparing the loss between the reconstructed image 1 and the reconstructed image 2, to ensure that the outputs of the reconstructed image 1 and the reconstructed image 2 are consistent;
  • Step 230 performing data consistency loss design on transformed image 1 generated after reconstructing image 1 and transformed image 2 generated after reconstructing image 2;
  • step 240 the trained model is tested, and the average value of the output images of the parallel network is used as the final reconstruction result.
  • the method in this application first performs subsampling on the undersampled data twice to obtain two data enhancement subsets. Combining the sensitivity maps of the two subsets of images produces the input image for the network.
  • a parallel network framework is built based on two data subsets, and the reconstructed network part can use the model-based MRI reconstruction method or other MRI reconstruction methods.
  • the reconstruction loss is mainly used to constrain the consistency between the output and input of the network to ensure that the network can better learn the mapping relationship between input and output.
  • Contrastive loss is mainly used to ensure consistency between the outputs of parallel networks, and to make better use of under-sampled data by virtue of the advantages of contrastive learning.
  • Data consistency loss is a further constraint on network training, the main purpose is to improve the quality of reconstructed images.
  • the model testing part the trained model is used for testing, and the average value of the output image of the parallel network is used as the final reconstruction result.
  • FIG. 3 shows an exemplary structural block diagram of an imaging device 300 for physics-based data-augmented contrast learning representation according to an embodiment of the present application.
  • the device includes:
  • An enhancement unit 310 configured to perform multi-physics undersampling data enhancement on the magnetic resonance data
  • a reconstruction unit 320 configured to utilize a contrastive representation learning framework for the enhanced undersampled magnetic resonance data
  • the guarantee unit 330 is configured to guarantee the accuracy of the magnetic resonance imaging by utilizing contrast loss constraints and data fitting items.
  • the present invention proposes a parallel network framework for fast magnetic resonance imaging, which provides guidance for magnetic resonance imaging technology.
  • FIG. 4 shows an exemplary structural block diagram of an imaging device 400 for physically-based data-enhanced contrastive learning representation according to another embodiment of the present application.
  • the device includes:
  • a reconstruction loss unit 410 configured to perform reconstruction loss design for the reconstructed image 1 and the reconstructed image 2;
  • the comparison loss unit 420 is used to compare the loss between the reconstructed image 1 and the reconstructed image 2, so as to ensure that the outputs of the reconstructed image 1 and the reconstructed image 2 are consistent;
  • the data consistency loss unit 430 is used to perform data consistency loss design on the transformed image 1 generated after the reconstructed image 1 and the transformed image 2 generated after the reconstructed image 2;
  • the model testing unit 440 is configured to test the trained model, and use the average value of the output images of the parallel network as the final reconstruction result.
  • the units or modules recorded in the apparatuses 300-400 correspond to the steps in the method described with reference to FIGS. 1-2. Therefore, the operations and features described above for the method are also applicable to the devices 300-400 and the units contained therein, and will not be repeated here.
  • the apparatuses 300-400 may be pre-implemented in the browser of the electronic device or other security applications, and may also be loaded into the browser of the electronic device or its security application by downloading or other means.
  • the corresponding units in the devices 300-400 may cooperate with the units in the electronic device to implement the solutions of the embodiments of the present application.
  • FIG. 5 shows a schematic structural diagram of a computer system 500 suitable for implementing a terminal device or a server according to an embodiment of the present application.
  • a computer system 500 includes a central processing unit (CPU) 501 that can be programmed according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage section 508 into a random-access memory (RAM) 503 Instead, various appropriate actions and processes are performed.
  • ROM read-only memory
  • RAM random-access memory
  • various programs and data required for the operation of the system 500 are also stored.
  • the CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504.
  • An input/output (I/O) interface 505 is also connected to the bus 504 .
  • the following components are connected to the I/O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 508 including a hard disk, etc. and a communication section 509 including a network interface card such as a LAN card, a modem, or the like.
  • the communication section 509 performs communication processing via a network such as the Internet.
  • a drive 510 is also connected to the I/O interface 505 as needed.
  • a removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 510 as necessary so that a computer program read therefrom is installed into the storage section 508 as necessary.
  • embodiments of the present disclosure include a method of physics-based data-augmented contrastive learning representation imaging comprising a computer program tangibly embodied on a machine-readable medium, the computer program comprising a method for performing the method of FIGS. 1-2 program code.
  • the computer program may be downloaded and installed from a network via communication portion 509, and/or installed from removable media 511.
  • each block in a flowchart or block diagram may represent a module, program segment, or part of code that includes one or more logical functions for implementing specified executable instructions.
  • the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations can be implemented by a dedicated hardware-based system that performs the specified functions or operations , or may be implemented by a combination of dedicated hardware and computer instructions.
  • the units or modules involved in the embodiments described in the present application may be implemented by means of software or by means of hardware.
  • the described units or modules may also be set in a processor.
  • a processor includes a first sub-region generating unit, a second sub-region generating unit, and a display region generating unit.
  • the names of these units or modules do not constitute limitations on the units or modules themselves in some cases, for example, the display area generation unit can also be described as "used to generate The cell of the display area of the text".
  • the present application also provides a computer-readable storage medium, which may be the computer-readable storage medium contained in the aforementioned devices in the above-mentioned embodiments; computer-readable storage media stored in the device.
  • the computer-readable storage medium stores one or more programs, and the aforementioned programs are used by one or more processors to execute the text generation method applied to transparent window envelopes described in this application.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Molecular Biology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Radiology & Medical Imaging (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Pathology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

本申请公开了一种基于物理的数据增强对比学习表示成像方法、装置、设备及其存储介质，该方法包括：对磁共振数据进行多物理轨迹的欠采样数据增强；对增强后的欠采样磁共振数据利用对比表示学习框架；利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。本申请提供的上述方案，解决了对于全采样数据的依赖问题，提高了对欠采样数据的利用效率。此外，本发明提出了一种磁共振快速成像的并行网络框架，为磁共振成像技术提供了指导。

Description

基于物理的数据增强对比学习表示成像方法 技术领域
本发明涉及医学影像技术，具体涉及一种基于物理的数据增强对比学习表示成像方法、装置、设备及其存储介质。
背景技术
磁共振成像(MRI)广泛应用于各种疾病的诊断，与计算机断层扫描(CT)等其他医学成像技术相比，MRI的一个瓶颈是扫描时间过长。并行MRI是一种广泛使用的加速策略，利用多个接收线圈提供的冗余信息来减少扫描时间。
近年来，深度学习以其优异的性能被广泛应用于MR图像重建。对于纯数据驱动的方法，通常需要学习欠采样数据和全采样数据之间的映射，或者有伪影和没有伪影图像之间的映射。为了提高方法的可解释性，提出了物理引导的方法，通常利用正则化最小二乘目标函数求解逆问题。这些模型主要采用全监督学习方法，在模型优化时需要全采样数据作为参考。尽管取得了成功，但在实际应用中很难获得完整的采样数据，尤其是在动态成像中。自监督学习方法可以解决这一问题，并取得了鼓舞人心的成绩。但是，由于收集的数据利用率较低，性能仍然可以得到改善。
现有的MRI重建方法的主要缺点是基于监督学习的方法过度依赖于全采样数据，在实际场景中，全采样数据很难获得或获得的成本 较大。而且长时间的扫描会引起患者的不适，并且对于心脏部分的成像会产生伪影。基于自监督的方法解决了对于全采样数据依赖的问题，但没有充分欠采样数据。
发明内容
鉴于现有技术中的上述缺陷或不足，期望提供一种基于物理的数据增强对比学习表示成像方法、装置、设备及其存储介质。
第一方面，本申请实施例提供了一种基于物理的数据增强对比学习表示成像方法，该方法包括：对磁共振数据进行多物理轨迹的欠采样数据增强；对增强后的欠采样磁共振数据利用对比表示学习框架；利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
在其中一个实施例中，所述对磁共振数据进行多物理轨迹的欠采样数据增强，包括：对欠采样数据进行二次欠采样得到数据增强子集一和数据增强子集二，将数据增强子集一和数据增强子集二结合图像的灵敏度图生成网络的输入图像。
在其中一个实施例中，所述对增强后的欠采样磁共振数据利用对比表示学习框架，包括：基于数据增强子集一和数据增强子集二搭建对应的并行网络框架重建网络一和重建网络二；将数据增强子集一输入到重建网络中生成重建图像一，将数据增强子集二输入到重建网络二中生成重建图像二。
在其中一个实施例中，所述利用对比损失约束以及数据拟合项，保证磁共振成像的准确性，包括：对于重建图像一和重建图像二进行重建损失设计；对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终 的重建结果。
第二方面，本申请实施例还提供了一种基于物理的数据增强对比学习表示成像装置，该装置包括：增强单元，用于对磁共振数据进行多物理轨迹的欠采样数据增强；重建单元，用于对增强后的欠采样磁共振数据利用对比表示学习框架；保证单元，用于利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
在其中一个实施例中，所述对磁共振数据进行多物理轨迹的欠采样数据增强，包括：对欠采样数据进行二次欠采样得到数据增强子集一和数据增强子集二，将数据增强子集一和数据增强子集二结合图像的灵敏度图生成网络的输入图像。
在其中一个实施例中，所述对增强后的欠采样磁共振数据利用对比表示学习框架，包括：基于数据增强子集一和数据增强子集二搭建对应的并行网络框架重建网络一和重建网络二；将数据增强子集一输入到重建网络中生成重建图像一，将数据增强子集二输入到重建网络二中生成重建图像二。
在其中一个实施例中，所述利用对比损失约束以及数据拟合项，保证磁共振成像的准确性，包括：重建损失单元，用于对于重建图像一和重建图像二进行重建损失设计；对比损失单元，用于对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；数据一致性损失单元，用于对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；模型测试单元，用于对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终的重建结果。
第三方面，本申请实施例还提供了一种计算机设备，包括存储器、处理器以及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如本申请实施例描述中任一所述的方法。
第四方面，本申请实施例还提供了一种计算机设备一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序用于：所述计算机程序被处理器执行时实现如本申请实施例描述中任一所述的方法。
本发明的有益效果：
本发明提供的基于物理的数据增强对比学习表示成像方法，解决了对于全采样数据的依赖问题，提高了对欠采样数据的利用效率。此外，本发明提出了一种磁共振快速成像的并行网络框架，为磁共振成像技术提供了指导。
附图说明
通过阅读参照以下附图所作的对非限制性实施例所作的详细描述，本申请的其它特征、目的和优点将会变得更明显：
图1示出了本申请实施例提供的基于物理的数据增强对比学习表示成像方法的流程示意图；
图2示出了本申请又一实施例提供的基于物理的数据增强对比学习表示成像方法的流程示意图；
图3示出了根据本申请一个实施例的基于物理的数据增强对比学习表示成像装置300的示例性结构框图；
图4示出了本申请又一实施例的基于物理的数据增强对比学习表示成像装置400的示例性结构框图；
图5示出了适于用来实现本申请实施例的终端设备的计算机系统的结构示意图；
图6示出了本申请实施例提供的又一流程图。
具体实施方式
为使本发明的上述目的、特征和优点能够更加明显易懂，下面结合附图对本发明的具体实施方式做详细的说明。在下面的描述中阐述了很多具体细节以便于充分理解本发明。但是本发明能够以很多不同于在此描述的其它方式来实施，本领域技术人员可以在不违背本发明内涵的情况下做类似改进，因此本发明不受下面公开的具体实施例的限制。
在本发明的描述中，需要理解的是，术语“中心”、“纵向”、“横向”、“长度”、“宽度”、“厚度”、“上”、“下”、“前”、“后”、“左”、“右”、“竖直”、“水平”、“顶”、“底”、“内”、“外”、“顺时针”、“逆时针”、“轴向”、“径向”、“周向”等指示的方位或位置关系为基于附图所示的方位或位置关系，仅是为了便于描述本发明和简化描述，而不是指示或暗示所指的装置或元件必须具有特定的方位、以特定的方位构造和操作，因此不能理解为对本发明的限制。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。在本发明的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。
在本发明中，除非另有明确的规定和限定，术语“安装”、“相连”、“连接”、“固定”等术语应做广义理解，例如，可以是固定连接，也可以是可拆卸连接，或成一体；可以是机械连接，也可以是电连接；可以是直接相连，也可以通过中间媒介间接相连，可以是两个元件内部的连通或两个元件的相互作用关系，除非另有明确的限定。对于本领域的普通技术人员而言，可以根据具体情况理解上述术语在本发明中的具体含义。
在本发明中，除非另有明确的规定和限定，第一特征在第二特征“上”或“下”可以是第一和第二特征直接接触，或第一和第二特征通过中间媒介间接接触。而且，第一特征在第二特征“之上”、“上方”和“上面”可是第一特征在第二特征正上方或斜上方，或仅仅表示第一特征水平高度高于第二特征。第一特征在第二特征“之下”、“下方”和“下面”可以是第一特征在第二特征正下方或斜下方，或仅仅表示第一特征水平高度小于第二特征。
需要说明的是，当元件被称为“固定于”或“设置于”另一个元件，它可以直接在另一个元件上或者也可以存在居中的元件。当一个元件被认为是“连接”另一个元件，它可以是直接连接到另一个元件或者可能同时存在居中元件。本文所使用的术语“垂直的”、“水平的”、“上”、“下”、“左”、“右”以及类似的表述只是为了说明的目的，并不表示是唯一的实施方式。
请参考图1，图1示出了本申请实施例提供的基于物理的数据增强对比学习表示成像方法的流程示意图。
如图1所示，该方法包括：
步骤110，对磁共振数据进行多物理轨迹的欠采样数据增强；
步骤120，对增强后的欠采样磁共振数据利用对比表示学习框架；
步骤130，利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
采用上述技术方案，解决了对于全采样数据的依赖问题，提高了对欠采样数据的利用效率。此外，本发明提出了一种磁共振快速成像的并行网络框架，为磁共振成像技术提供了指导。
在一些实施例中，参考图6所示，本申请中的对磁共振数据进行多物理轨迹的欠采样数据增强，包括：对欠采样数据进行二次欠采样得到数据增强子集一和数据增强子集二，将数据增强子集一和数据增 强子集二结合图像的灵敏度图生成网络的输入图像。
在一些实施例中，参考图6所示，本申请中的对增强后的欠采样磁共振数据利用对比表示学习框架，包括：基于数据增强子集一和数据增强子集二搭建对应的并行网络框架重建网络一和重建网络二；将数据增强子集一输入到重建网络中生成重建图像一，将数据增强子集二输入到重建网络二中生成重建图像二。
在一些实施例中，请参考图2，图2给出了本申请又一实施例提供的基于物理的数据增强对比学习表示成像方法的流程示意图。
如图2所示，其中利用对比损失约束以及数据拟合项，保证磁共振成像的准确性，包括：
步骤210，对于重建图像一和重建图像二进行重建损失设计；
步骤220，对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；
步骤230，对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；
步骤240，对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终的重建结果。
综上所述，参考图6所示，本申请中的方法首先将欠采样数据进行二次欠采样得到两个数据增强子集。将两个子集结合图像的灵敏度图生成网络的输入图像。基于两个数据子集搭建一个并行网络框架，其中重建网络部分可以使用基于模型的MRI重建方法或者其他的MRI重建方法。重建损失主要是用来约束网络的输出和输入之间的一致性，保证网络更好的学习输入和输出之间的映射关系。对比损失主要用来保证并行网络的输出之间保持一致，借助于对比学习的优势更好的利用欠采样数据。数据一致性损失是对网络训练的进一步约束，主要目的提高重建图像的质量。模型测试部分是利用训练好的模型进 行测试，将并行网络的输出图像的平均值作为最后的重建结果。
进一步地，参考图3，图3示出了根据本申请一个实施例的基于物理的数据增强对比学习表示成像装置300的示例性结构框图。
如图3所示，该装置包括：
增强单元310，用于对磁共振数据进行多物理轨迹的欠采样数据增强；
重建单元320，用于对增强后的欠采样磁共振数据利用对比表示学习框架；
保证单元330，用于利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
采用上述技术方案，解决了对于全采样数据的依赖问题，提高了对欠采样数据的利用效率。此外，本发明提出了一种磁共振快速成像的并行网络框架，为磁共振成像技术提供了指导。
进一步地，请参考图4，图4给出了本申请又一实施例的基于物理的数据增强对比学习表示成像装置400的示例性结构框图。
如图4所示，该装置包括：
重建损失单元410，用于对于重建图像一和重建图像二进行重建损失设计；
对比损失单元420，用于对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；
数据一致性损失单元430，用于对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；
模型测试单元440，用于对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终的重建结果。
应当理解，装置300-400中记载的诸单元或模块与参考图1-2描述的方法中的各个步骤相对应。由此，上文针对方法描述的操作和特 征同样适用于装置300-400及其中包含的单元，在此不再赘述。装置300-400可以预先实现在电子设备的浏览器或其他安全应用中，也可以通过下载等方式而加载到电子设备的浏览器或其安全应用中。装置300-400中的相应单元可以与电子设备中的单元相互配合以实现本申请实施例的方案。
下面参考图5，其示出了适于用来实现本申请实施例的终端设备或服务器的计算机系统500的结构示意图。
如图5所示，计算机系统500包括中央处理单元(CPU)501，其可以根据存储在只读存储器(ROM)502中的程序或者从存储部分508加载到随机访问存储器(RAM)503中的程序而执行各种适当的动作和处理。在RAM 503中，还存储有系统500操作所需的各种程序和数据。CPU 501、ROM 502以及RAM 503通过总线504彼此相连。输入/输出(I/O)接口505也连接至总线504。
以下部件连接至I/O接口505：包括键盘、鼠标等的输入部分506；包括诸如阴极射线管(CRT)、液晶显示器(LCD)等以及扬声器等的输出部分507；包括硬盘等的存储部分508；以及包括诸如LAN卡、调制解调器等的网络接口卡的通信部分509。通信部分509经由诸如因特网的网络执行通信处理。驱动器510也根据需要连接至I/O接口505。可拆卸介质511，诸如磁盘、光盘、磁光盘、半导体存储器等等，根据需要安装在驱动器510上，以便于从其上读出的计算机程序根据需要被安装入存储部分508。
特别地，根据本公开的实施例，上文参考图1-2描述的过程可以被实现为计算机软件程序。例如，本公开的实施例包括一种基于物理的数据增强对比学习表示成像方法，其包括有形地包含在机器可读介质上的计算机程序，所述计算机程序包含用于执行图1-2的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信部分509从 网络上被下载和安装，和/或从可拆卸介质511被安装。
附图中的流程图和框图，图示了按照本发明各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段、或代码的一部分，前述模块、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
描述于本申请实施例中所涉及到的单元或模块可以通过软件的方式实现，也可以通过硬件的方式来实现。所描述的单元或模块也可以设置在处理器中，例如，可以描述为：一种处理器包括第一子区域生成单元、第二子区域生成单元以及显示区域生成单元。其中，这些单元或模块的名称在某种情况下并不构成对该单元或模块本身的限定，例如，显示区域生成单元还可以被描述为“用于根据第一子区域和第二子区域生成文本的显示区域的单元”。
作为另一方面，本申请还提供了一种计算机可读存储介质，该计算机可读存储介质可以是上述实施例中前述装置中所包含的计算机可读存储介质；也可以是单独存在，未装配入设备中的计算机可读存储介质。计算机可读存储介质存储有一个或者一个以上程序，前述程序被一个或者一个以上的处理器用来执行描述于本申请的应用于透明窗口信封的文本生成方法。
以上描述仅为本申请的较佳实施例以及对所运用技术原理的说明。 本领域技术人员应当理解，本申请中所涉及的发明范围，并不限于上述技术特征的特定组合而成的技术方案，同时也应涵盖在不脱离前述发明构思的情况下，由上述技术特征或其等同特征进行任意组合而形成的其它技术方案。例如上述特征与本申请中公开的(但不限于)具有类似功能的技术特征进行互相替换而形成的技术方案。

Claims (10)

  1. 一种基于物理的数据增强对比学习表示成像方法，其特征在于，该方法包括：
    对磁共振数据进行多物理轨迹的欠采样数据增强；
    对增强后的欠采样磁共振数据利用对比表示学习框架；
    利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
  2. 根据权利要求1所述的基于物理的数据增强对比学习表示成像方法，其特征在于，所述对磁共振数据进行多物理轨迹的欠采样数据增强，包括：
    对欠采样数据进行二次欠采样得到数据增强子集一和数据增强子集二，将数据增强子集一和数据增强子集二结合图像的灵敏度图生成网络的输入图像。
  3. 根据权利要求2所述的基于物理的数据增强对比学习表示成像方法，其特征在于，所述对增强后的欠采样磁共振数据利用对比表示学习框架，包括：
    基于数据增强子集一和数据增强子集二搭建对应的并行网络框架重建网络一和重建网络二；
    将数据增强子集一输入到重建网络中生成重建图像一，将数据增强子集二输入到重建网络二中生成重建图像二。
  4. 根据权利要求3所述的基于物理的数据增强对比学习表示成像方法，其特征在于，所述利用对比损失约束以及数据拟合项，保证磁共振成像的准确性，包括：
    对于重建图像一和重建图像二进行重建损失设计；
    对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；
    对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；
    对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终的重建结果。
  5. 一种基于物理的数据增强对比学习表示成像装置，其特征在于，该装置包括：
    增强单元，用于对磁共振数据进行多物理轨迹的欠采样数据增强；
    重建单元，用于对增强后的欠采样磁共振数据利用对比表示学习框架；
    保证单元，用于利用对比损失约束以及数据拟合项，保证磁共振成像的准确性。
  6. 根据权利要求5所述的基于物理的数据增强对比学习表示成像装置，其特征在于，所述对磁共振数据进行多物理轨迹的欠采样数据增强，包括：
    对欠采样数据进行二次欠采样得到数据增强子集一和数据增强子集二，将数据增强子集一和数据增强子集二结合图像的灵敏度图生成网络的输入图像。
  7. 根据权利要求6所述的基于物理的数据增强对比学习表示成像装置，其特征在于，所述对增强后的欠采样磁共振数据利用对比表示学习框架，包括：
    基于数据增强子集一和数据增强子集二搭建对应的并行网络框架重建网络一和重建网络二；
    将数据增强子集一输入到重建网络中生成重建图像一，将数据增强子集二输入到重建网络二中生成重建图像二。
  8. 根据权利要求7所述的基于物理的数据增强对比学习表示成像装置，其特征在于，所述利用对比损失约束以及数据拟合项，保证磁 共振成像的准确性，包括：
    重建损失单元，用于对于重建图像一和重建图像二进行重建损失设计；
    对比损失单元，用于对比重建图像一和重建图像二之间的损失，保证重建图像一和重建图像二的输出保持一致；
    数据一致性损失单元，用于对重建图像一之后生成的变换图像一和重建图像二之后生成的变换图像二进行数据一致性损失设计；
    模型测试单元，用于对训练好的模型进行测试，将并行网络的输出图像的平均值作为最终的重建结果。
  9. 一种计算机设备，包括存储器、处理器以及存储在存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如权利要求1-4中任一所述的方法。
  10. 一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序用于：
    所述计算机程序被处理器执行时实现如权利要求1-4中任一所述的方法。
PCT/CN2021/140119 2021-12-21 2021-12-21 基于物理的数据增强对比学习表示成像方法 Ceased WO2023115339A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/140119 WO2023115339A1 (zh) 2021-12-21 2021-12-21 基于物理的数据增强对比学习表示成像方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/140119 WO2023115339A1 (zh) 2021-12-21 2021-12-21 基于物理的数据增强对比学习表示成像方法

Publications (1)

Publication Number Publication Date
WO2023115339A1 true WO2023115339A1 (zh) 2023-06-29

Family

ID=86900946

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/140119 Ceased WO2023115339A1 (zh) 2021-12-21 2021-12-21 基于物理的数据增强对比学习表示成像方法

Country Status (1)

Country Link
WO (1) WO2023115339A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106491131A (zh) * 2016-12-30 2017-03-15 深圳先进技术研究院 一种磁共振的动态成像方法和装置
CN110992440A (zh) * 2019-12-10 2020-04-10 中国科学院深圳先进技术研究院 弱监督磁共振快速成像方法和装置
CN113192151A (zh) * 2021-04-08 2021-07-30 广东工业大学 一种基于结构相似性的mri图像重建方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106491131A (zh) * 2016-12-30 2017-03-15 深圳先进技术研究院 一种磁共振的动态成像方法和装置
CN110992440A (zh) * 2019-12-10 2020-04-10 中国科学院深圳先进技术研究院 弱监督磁共振快速成像方法和装置
CN113192151A (zh) * 2021-04-08 2021-07-30 广东工业大学 一种基于结构相似性的mri图像重建方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
CHENG HUI-TAO, WANG SHAN-SHAN, KE ZI-WEN, JIA SEN, CHENG JING, QIU ZHI-LANG, ZHENG HAI-RONG, LIANG DONG: "A Deep Recursive Cascaded Convolutional Network for Parallel MRI", CHINESE JOURNAL OF MAGNETIC RESONANCE, vol. 36, no. 4, 15 December 2019 (2019-12-15), CN , pages 437 - 445, XP093068441, ISSN: 1000-4556, DOI: 10.11938/cjmr20192721 *

Similar Documents

Publication Publication Date Title
CN109961491B (zh) 多模态图像截断补偿方法、装置、计算机设备和介质
CN110249365B (zh) 用于图像重建的系统和方法
Liu et al. Learning MRI artefact removal with unpaired data
Yushkevich et al. Bias in estimation of hippocampal atrophy using deformation-based morphometry arises from asymmetric global normalization: an illustration in ADNI 3 T MRI data
US20170372193A1 (en) Image Correction Using A Deep Generative Machine-Learning Model
Shen et al. Rapid reconstruction of highly undersampled, non‐Cartesian real‐time cine k‐space data using a perceptual complex neural network (PCNN)
CN111243052B (zh) 图像重建方法、装置、计算机设备和存储介质
CN115115723B (zh) 图像重建模型生成及图像重建方法、装置、设备和介质
CN115115722B (zh) 图像重建模型生成及图像重建方法、装置、设备和介质
CN112365560A (zh) 基于多级网络的图像重建方法、系统、可读存储介质和设备
CN111714124A (zh) 磁共振电影成像方法、装置、成像设备及存储介质
Van Der Geest et al. Evaluation of a new method for automated detection of left ventricular boundaries in time series of magnetic resonance images using an active appearance motion model
WO2021120069A1 (zh) 基于解剖结构差异先验的低剂量图像重建方法和系统
US20230079353A1 (en) Image correction using an invertable network
Schultz et al. HiFiVE: a Hilbert space embedding of fiber variability estimates for uncertainty modeling and visualization
CN108010094A (zh) 一种磁共振图像重建方法和装置
CN113763399B (zh) 一种基于弱监督学习的医学影像分割方法及计算机可读存储介质
WO2024093083A1 (zh) 一种基于变分自编码器的磁共振加权图像合成方法和装置
Guan et al. Magnetic resonance imaging reconstruction using a deep energy‐based model
San José Estépar Artificial intelligence in functional imaging of the lung
CN103720475A (zh) 医学图像成像方法及使用所述方法的医学诊断设备
Rai et al. Deep Learning–Based Acceleration in MRI: Current Landscape and Clinical Applications in Neuroradiology
CN111681297A (zh) 图像重建方法、计算机设备和存储介质
Huemer et al. Improved quantification in CEST‐MRI by joint spatial total generalized variation
Marinelli et al. Automatic PET‐CT Image Registration Method Based on Mutual Information and Genetic Algorithms

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: 21968492

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21968492

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 21968492

Country of ref document: EP

Kind code of ref document: A1

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 20/01/2025)