WO2022099454A1

WO2022099454A1 - Image segmentation method, terminal device, and computer-readable storage medium

Info

Publication number: WO2022099454A1
Application number: PCT/CN2020/127785
Authority: WO
Inventors: 邹超; 程传力; 王志明; 刘新; 郑海荣
Original assignee: 中国科学院深圳先进技术研究院
Priority date: 2020-11-10
Filing date: 2020-11-10
Publication date: 2022-05-19

Abstract

The present application is applicable to the technical field of image processing, and provides an image segmentation method, a terminal device, and a computer-readable storage medium. A proton density fat fraction quantitative image of whole-body fat distribution that is used for accurate quantification is obtained, then the proton density fat fraction quantitative image of the whole-body fat distribution is processed by means of a preset image segmentation model, and thus accurate segmentation of fat in parts of a human body is implemented. The problems that existing fat quantification and segmentation methods are usually used to process a certain part and cannot achieve segmentation of whole-body fat, quantification of deposition degrees of whole-body fat, and accurate segmentation of whole-body fat, are solved.

Description

一种图像分割方法、终端设备及计算机可读存储介质Image segmentation method, terminal device and computer-readable storage medium

技术领域technical field

本申请属于图像处理技术领域，尤其涉及一种图像分割方法、终端设备及计算机可读存储介质。The present application belongs to the technical field of image processing, and in particular, relates to an image segmentation method, a terminal device and a computer-readable storage medium.

背景技术Background technique

肥胖是由于体内脂肪组织过度积累造成的，肥胖有可能会导致多种慢性疾病的发生。为了研究肥胖机理和探索抵抗肥胖的方法，就需要对体内脂肪进行精准定量分析和分割。然而，现有的脂肪定量和分割方法通常是针对某一部位进行处理，无法实现全身脂肪的分割和全身脂肪沉积程度的定量和全身脂肪的准确分割。Obesity is caused by excessive accumulation of adipose tissue in the body, and obesity may lead to the occurrence of various chronic diseases. In order to study the mechanism of obesity and explore methods to resist obesity, it is necessary to carry out accurate quantitative analysis and segmentation of body fat. However, the existing fat quantification and segmentation methods usually deal with a certain part, and cannot achieve the segmentation of whole body fat, the quantification of the degree of whole body fat deposition and the accurate segmentation of whole body fat.

技术问题technical problem

有鉴于此，本申请实施例提供了一种图像分割方法、终端设备及计算机可读存储介质，以解决解决了现有的脂肪定量和分割方法通常是针对某一部位进行处理，无法实现全身脂肪的分割和全身脂肪沉积程度的定量和全身脂肪的准确分割的问题。In view of this, the embodiments of the present application provide an image segmentation method, a terminal device, and a computer-readable storage medium, so as to solve the problem that the existing fat quantification and segmentation methods usually deal with a certain part, and cannot achieve whole body fat. The segmentation and quantification of the extent of whole body fat deposition and the problem of accurate segmentation of whole body fat.

技术解决方案technical solutions

第一方面，本申请实施例提供一种图像分割方法，包括：In a first aspect, an embodiment of the present application provides an image segmentation method, including:

获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；Acquiring a target image; the target image is a proton density fat fraction quantitative map of the body fat distribution of the subject;

将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The target image is input into a preset image segmentation model for processing, and subcutaneous fat images and visceral fat images of each preset part are segmented from the target image.

可选的，上述获取目标图像的步骤包括：Optionally, the above step of acquiring the target image includes:

获取多个不同回波时间的磁共振图像；Acquiring multiple magnetic resonance images with different echo times;

根据所述多个不同回波时间的磁共振图像确定目标图像。The target image is determined from the plurality of magnetic resonance images with different echo times.

可选的，所述将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图，包括：Optionally, the target image is input into a preset image segmentation model for processing, and subcutaneous fat images and visceral fat maps of each preset part are segmented from the target image, including:

从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；Identify the precise proton density fat fraction quantitative map corresponding to each preset part from the target image;

基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；Determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part.

可选的，所述根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：Optionally, the subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part, including:

对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；performing a downsampling operation on the overall fat map corresponding to each preset part, and extracting edge features of the total fat map corresponding to each preset part to obtain an edge feature map;

对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图。An up-sampling operation is performed on the edge feature map, the spatial information and edge information in the overall fat map corresponding to each preset part are restored, and a subcutaneous fat map and a visceral fat map corresponding to each preset part are obtained.

可选的，所述对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图，包括：Optionally, performing a down-sampling operation on the overall fat map corresponding to each preset part, and extracting edge features of the overall fat map corresponding to each preset part to obtain an edge feature map, including:

对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；Perform down-sampling processing on the overall fat map corresponding to each preset part to obtain a first down-sampling feature map;

对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；performing down-sampling processing on the first down-sampling feature map to obtain a second down-sampling feature map;

对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；performing down-sampling processing on the second down-sampling feature map to obtain a third down-sampling feature map;

对所述第三下采样特征图进行下采样处理，得到边缘特征图。Perform down-sampling processing on the third down-sampling feature map to obtain an edge feature map.

可选的，所述对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：Optionally, performing an upsampling operation on the edge feature map, recovering the spatial information and edge information in the overall fat map corresponding to each preset part, and obtaining the subcutaneous fat map and visceral fat map corresponding to each preset part. Figures, including:

对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第一通道特征图进行融合，得到第一融合特征图；Perform up-sampling processing on the edge feature map to obtain a first up-sampling feature map, and fuse the first up-sampling feature map with the first channel feature map to obtain a first fused feature map;

对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第二通道特征图进行融合，得到第二融合特征图；Perform up-sampling processing on the first fusion feature map to obtain a second up-sampling feature map, and fuse the second sampling feature map with the second channel feature map to obtain a second fusion feature map;

对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第三通道特征图进行融合，得到第三融合特征图；Perform upsampling processing on the second fusion feature map to obtain a third upsampling feature map, and fuse the third upsampling feature map with the third channel feature map to obtain a third fusion feature map;

对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第四通道特征图进行融合，得到分割结果图。Perform up-sampling processing on the third fusion feature map to obtain a fourth up-sampling feature map, and fuse the fourth up-sampling feature map with the fourth channel feature map to obtain a segmentation result map.

可选的，上述图像分割方法，还包括：Optionally, the above image segmentation method further includes:

构建图像分割模型；Build an image segmentation model;

获取样本数据集；其中，所述样本数据集包括多组样本数据，每组样本数据包括样本图像和标签图像，所述样本图像是指覆盖人体颈部到膝盖部位的质子密度脂肪分数定量图，所述标签图像是指对应于质子密度脂肪分数定量图的皮下脂肪图和内脏脂肪图；Obtaining a sample data set; wherein, the sample data set includes multiple sets of sample data, each set of sample data includes a sample image and a label image, and the sample image refers to a quantitative map of proton density fat fraction covering the part of the human body from the neck to the knee, The label image refers to a subcutaneous fat map and a visceral fat map corresponding to a quantitative map of proton density fat fraction;

基于样本数据集对所述预先构建的图像分割模型进行训练，得到所述预设图像分割模型。The pre-built image segmentation model is trained based on the sample data set to obtain the preset image segmentation model.

第二方面，本申请实施例提供一种终端设备，包括：In a second aspect, an embodiment of the present application provides a terminal device, including:

第一获取单元，用于获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；The first acquisition unit is used to acquire a target image; the target image is a proton density fat fraction quantitative map of the whole body fat distribution of the measured object;

第一处理单元，用于将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The first processing unit is configured to input the target image into a preset image segmentation model for processing, and segment the subcutaneous fat image and the visceral fat image of each preset part from the target image.

第三方面，本申请实施例提供一种终端设备，所述终端设备包括处理器、存储器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现如第一方面或第一方面的任意可选方式所述的方法。In a third aspect, an embodiment of the present application provides a terminal device, where the terminal device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, where the processor executes the A computer program implements the method as described in the first aspect or any alternative of the first aspect.

第四方面，本申请实施例提供一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行时实现如第一方面或第一方面的任意可选方式所述的方法。In a fourth aspect, embodiments of the present application provide a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, implements the first aspect or any of the first aspect. Select the method described in the method.

第五方面，本申请实施例提供一种计算机程序产品，当计算机程序产品在终端设备上运行时，使得终端设备执行上述第一方面或第一方面的任意可选方式所述的方法。In a fifth aspect, an embodiment of the present application provides a computer program product that, when the computer program product runs on a terminal device, enables the terminal device to execute the method described in the first aspect or any optional manner of the first aspect.

有益效果beneficial effect

实施本申请实施例提供的一种图像分割方法、终端设备、计算机可读存储介质及计算机程序产品具有以下有益效果：Implementing an image segmentation method, a terminal device, a computer-readable storage medium, and a computer program product provided by the embodiments of the present application has the following beneficial effects:

通过获取能够精准定量的全身脂肪分布的质子密度脂肪分数定量图，再通过预设的图像分割模型对全身脂肪分布的质子密度脂肪分数定量图进行处理，进而实现对人体的各个部分的脂肪进行准确分割。解决了现有的脂肪定量和分割方法通常是针对某一部位进行处理，无法实现全身脂肪的分割和全身脂肪沉积程度的定量和全身脂肪的准确分割的问题。By obtaining the proton density fat fraction quantitative map of the whole body fat distribution that can be accurately quantified, and then processing the proton density fat fraction quantitative map of the whole body fat distribution through the preset image segmentation model, so as to realize the accurate analysis of the fat of each part of the human body. segmentation. It solves the problem that the existing fat quantification and segmentation methods usually deal with a certain part, and cannot achieve the segmentation of whole body fat, the quantification of the degree of whole body fat deposition and the accurate segmentation of whole body fat.

附图说明Description of drawings

为了更清楚地说明本申请实施例中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅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性的前提下，还可以根据这些附图获得其他的附图。In order to illustrate the technical solutions in the embodiments of the present application more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only for the present application. In some embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort.

图1是本申请实施例提供的一种图像分割方法的示意性流程图；FIG. 1 is a schematic flowchart of an image segmentation method provided by an embodiment of the present application;

图2是本申请实施例提供的一种图像分割方法中的实现过程示意图；2 is a schematic diagram of an implementation process in an image segmentation method provided by an embodiment of the present application;

图3是本申请实施例提供的一种图像分割模型的结构示意图；3 is a schematic structural diagram of an image segmentation model provided by an embodiment of the present application;

图4是本申请另一实施例提供的一种图像分割模型的结构示意图；4 is a schematic structural diagram of an image segmentation model provided by another embodiment of the present application;

图5是本申请实施例提供的一种终端设备的结构示意图；FIG. 5 is a schematic structural diagram of a terminal device provided by an embodiment of the present application;

图6是本申请另一实施例提供的一种终端设备的结构示意图；6 is a schematic structural diagram of a terminal device provided by another embodiment of the present application;

图7是本申请实施例提供的一种计算机可读存储介质的结构示意图。FIG. 7 is a schematic structural diagram of a computer-readable storage medium provided by an embodiment of the present application.

本发明的实施方式Embodiments of the present invention

以下描述中，为了说明而不是为了限定，提出了诸如特定***结构、技术之类的具体细节，以便透彻理解本申请实施例。然而，本领域的技术人员应当清楚，在没有这些具体细节的其它实施例中也可以实现本申请。在其它情况中，省略对众所周知的***、装置、电路以及方法的详细说明，以免不必要的细节妨碍本申请的描述。In the following description, for the purpose of illustration rather than limitation, specific details such as a specific system structure and technology are set forth in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be practiced in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

应当理解，在本申请说明书和所附权利要求书中使用的术语“和/或”是指相关联列出的项中的一个或多个的任何组合以及所有可能组合，并且包括这些组合。另外，在本申请说明书和所附权利要求书的描述中，术语“第一”、“第二”、“第三”等仅用于区分描述，而不能理解为指示或暗示相对重要性。It should be understood that, as used in this specification and the appended claims, the term "and/or" refers to and including any and all possible combinations of one or more of the associated listed items. In addition, in the description of the specification of the present application and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the description, and should not be construed as indicating or implying relative importance.

还应当理解，在本申请说明书中描述的参考“一个实施例”或“一些实施例”等意味着在本申请的一个或多个实施例中包括结合该实施例描述的特定特征、结构或特点。由此，在本说明书中的不同之处出现的语句“在一个实施例中”、“在一些实施例中”、“在其他一些实施例中”、“在另外一些实施例中”等不是必然都参考相同的实施例，而是意味着“一个或多个但不是所有的实施例”，除非是以其他方式另外特别强调。术语“包括”、“包含”、“具有”及它们的变形都意味着“包括但不限于”，除非是以其他方式另外特别强调。It should also be understood that references to "one embodiment" or "some embodiments" and the like described in the specification of this application mean that a particular feature, structure or characteristic described in connection with the embodiment is included in one or more embodiments of the application . Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in other embodiments," etc. in various places in this specification are not necessarily All refer to the same embodiment, but mean "one or more but not all embodiments" unless specifically emphasized otherwise. The terms "including", "including", "having" and their variants mean "including but not limited to" unless specifically emphasized otherwise.

人体的体内脂肪主要包括皮下脂肪和内脏脂肪。其中，皮下脂肪是指贮存在真皮层以下，深层筋膜层以上被浅筋膜包裹的储能细胞中的脂肪组织，主要用于人体保温及能量存储。内脏脂肪是指主要贮存于腹腔内的脂肪组织，内脏脂肪围绕着人的脏器，对人的内脏起着支撑、稳定和保护的作用。内脏脂肪的过量囤积有可能导致多种慢性疾病的发生，例如心血管疾病、糖尿病、脂肪肝等。因此，确定出人体的体内脂肪的分布和含量对于评估人体的健康状况有着极其重要的作用。Body fat mainly includes subcutaneous fat and visceral fat. Among them, subcutaneous fat refers to the adipose tissue stored in the energy storage cells below the dermis layer and above the deep fascia layer and wrapped by the superficial fascia, which is mainly used for human body heat preservation and energy storage. Visceral fat refers to the adipose tissue mainly stored in the abdominal cavity. Visceral fat surrounds human organs and plays a role in supporting, stabilizing and protecting human internal organs. Excessive accumulation of visceral fat may lead to the occurrence of various chronic diseases, such as cardiovascular disease, diabetes, fatty liver and so on. Therefore, determining the distribution and content of body fat in the human body plays an extremely important role in evaluating the health status of the human body.

随着医疗影像学设备的发展，基于医学图像去分析人体的健康状况已成为重要的分析手段。核磁共振成像技术以其多参数成像、无电离辐射等优点已经成为对人体脂肪定量分析的有力手段。目前基于核磁共振成像技术对人体脂肪进行定量和分割主要是基于组织纵向弛豫时间T1差异来实现脂肪的识别和分割，或者基于化学位移差异来实现脂肪的识别和分割。其中，对人体脂肪进行定量是指确定人体中脂肪的含量，分割是指将人体中的皮下脂肪和内脏脂肪进行区分。With the development of medical imaging equipment, the analysis of human health status based on medical images has become an important analysis method. Magnetic resonance imaging technology has become a powerful method for quantitative analysis of human body fat due to its advantages of multi-parameter imaging and no ionizing radiation. At present, the quantification and segmentation of human body fat based on MRI technology is mainly based on the difference of tissue longitudinal relaxation time T1 to realize fat identification and segmentation, or based on chemical shift difference to realize fat identification and segmentation. Among them, quantifying human body fat refers to determining the content of fat in the human body, and segmentation refers to distinguishing subcutaneous fat and visceral fat in the human body.

具体的，基于组织纵向弛豫时间T1差异的脂肪的识别和分割方法一般采用快速自旋回波(fast spin echo，FSE)成像，根据脂肪组织信号与周围组织信号的对比度，采用手动或半自动的分割方法来实现脂肪的分割，其中半自动分割方法主要是根据回波图像的灰度直方图分布，通过不断调整分割阈值实现脂肪组织与周围组织的区分。其中，组织纵向弛豫时间T1是指90射频脉冲纵向磁化矢量由零增长到它的最大值的63％所需要的时间。Specifically, the identification and segmentation method of fat based on the difference of tissue longitudinal relaxation time T1 generally adopts fast spin echo (FSE) imaging, and adopts manual or semi-automatic segmentation according to the contrast between adipose tissue signal and surrounding tissue signal. The semi-automatic segmentation method is mainly based on the gray histogram distribution of the echo image, and the adipose tissue is distinguished from the surrounding tissue by continuously adjusting the segmentation threshold. The tissue longitudinal relaxation time T1 refers to the time required for the longitudinal magnetization vector of 90 radio frequency pulses to increase from zero to 63% of its maximum value.

基于组织化学位移差异的成像方法也称为磁共振化学位移编码成像方法，磁共振化学位移编码成像方法利用水和脂肪的化学位移差异经过水脂信号分离得到脂肪含量定量分布图，在该分布图中，脂肪含量越高的区域，其定量数值越接近于100％，反之，脂肪含量越低的区域，其定量数值越接近于0，然后采用半自动或全自动的分割方法来实现脂肪组织的分割。常用的分割方法有地图集方法和卷积神经网络方法。地图集方法首先采用半自动分割得到脂肪分割图像作为地图集，然后将地图集向待分割图像进行配准，根据地图集中脂肪分布特征实现目标图像的脂肪分割。卷积神经网络方法通过对人体某一机体部位进行深度学习，进而实现脂肪分割。The imaging method based on histochemical shift difference is also called the magnetic resonance chemical shift coding imaging method. The magnetic resonance chemical shift coding imaging method uses the chemical shift difference of water and fat to obtain a quantitative distribution map of fat content through separation of water and fat signals. In the region with higher fat content, the quantitative value is closer to 100%, on the contrary, the quantitative value of the region with lower fat content is closer to 0, and then semi-automatic or fully automatic segmentation methods are used to achieve adipose tissue segmentation. . Commonly used segmentation methods include atlas method and convolutional neural network method. The atlas method first uses semi-automatic segmentation to obtain fat segmentation images as atlases, and then registers the atlases to the images to be segmented, and realizes fat segmentation of the target images according to the fat distribution characteristics of the atlases. The convolutional neural network method realizes fat segmentation through deep learning of a certain body part of the human body.

然而，基于组织纵向弛豫时间T1差异的脂肪识别和分割方法具有以下主要缺陷：1、组织对比度(即脂肪组织信号和周围组织信号的对比度)容易受到射频场的不均匀性和多通道线圈的灵敏度等因素的影响，造成对脂肪识别时的误差；2、手动或半自动分割方法费时费力；3、无法对内脏脂肪的沉积程度进行准确的定量。However, fat identification and segmentation methods based on differences in tissue longitudinal relaxation times, T1, have the following major drawbacks: 1. Tissue contrast (i.e., the contrast between adipose tissue signal and surrounding tissue signal) is susceptible to radio frequency field inhomogeneity and multi-channel coils. Sensitivity and other factors lead to errors in fat identification; 2. Manual or semi-automatic segmentation methods are time-consuming and labor-intensive; 3. The degree of visceral fat deposition cannot be accurately quantified.

而基于组织化学位移差异的的脂肪定量和分割方法存在的主要缺陷有：1、地图集方法流程复杂、自由参数多，造成分割结果稳定性不足；2、没有进行脏器分割和对内脏脂肪的沉积程度的准确定量；3、仅针对某一机体部位进行训练、分割，无法实现全身脂肪的精准定量和分割。The main defects of fat quantification and segmentation methods based on histochemical shift differences are: 1. The atlas method has a complex process and many free parameters, resulting in insufficient stability of the segmentation results; 2. There is no organ segmentation and visceral fat segmentation. Accurate quantification of the degree of deposition; 3. Only training and segmentation for a certain body part cannot achieve accurate quantification and segmentation of whole body fat.

为了解决上述缺陷，本申请实施例通过将被测对象的全身脂肪分布的质子密度脂肪分数定量图作为目标图像，再将目标图像输入到预设的图像分割模型中，通过预设的图像分割模型对该目标图像进行处理，进而实现对人体的各个部分的脂肪进行准确分割。以下将对本申请实施例提供的图像分割方法进行详细的说明：In order to solve the above-mentioned defects, the embodiment of the present application uses the proton density fat fraction quantitative map of the whole body fat distribution of the measured object as the target image, and then inputs the target image into the preset image segmentation model, through the preset image segmentation model The target image is processed to achieve accurate segmentation of the fat of each part of the human body. The image segmentation method provided by the embodiments of the present application will be described in detail below:

请参阅图1，图1是本申请实施例提供的一种图像分割方法的示意性流程图。本申请实施例提供的图像分割方法的执行主体为终端设备，终端设备可以是智能手机、平板电脑或可穿戴设备等移动终端，也可以是各种应用场景下的电脑、云服务器、医疗辅助计算机等。Please refer to FIG. 1. FIG. 1 is a schematic flowchart of an image segmentation method provided by an embodiment of the present application. The execution subject of the image segmentation method provided by the embodiments of the present application is a terminal device, and the terminal device may be a mobile terminal such as a smartphone, a tablet computer, or a wearable device, or a computer, a cloud server, or a medical assistant computer in various application scenarios. Wait.

如图1所示的图像分割方法可以包括S11～S12，详述如下：The image segmentation method shown in FIG. 1 may include S11 to S12, which are described in detail as follows:

S11：获取目标图像。S11: Acquire a target image.

本申请实施例中，上述目标图像是指能够体现被测对象的全身脂肪分布的质子密度脂肪分数(proton density fat fraction，PDFF)定量图。质子密度脂肪分数定量图能够精准地反映脂肪组织的占比。In the embodiment of the present application, the above-mentioned target image refers to a quantitative proton density fat fraction (PDFF) map that can reflect the whole body fat distribution of the measured object. The proton density fat fraction quantitative map can accurately reflect the proportion of adipose tissue.

在具体应用中，可以通过磁共振化学位移编码成像技术得到目标图像。In specific applications, the target image can be obtained through the magnetic resonance chemical shift coding imaging technology.

人体组织内水中的氢质子和脂肪中的氢质子存在化学位移，利用磁共振化学位移编码成像技术，使用磁共振梯度回波成像序列采集多个不同回波时间的磁共振图像，在不同回波时间下的磁共振图像中水氢质子与脂肪氢质子具有不同的相位差。再通过预设的编码成像模型求解，就可以将水氢质子和脂肪氢质子分离得到纯水图像和纯脂肪图像，再根据质子密度脂肪分数＝脂肪图像/(脂肪图像+水图像)可以得到质子密度脂肪分数。Hydrogen protons in water in human tissues and hydrogen protons in fat have chemical shifts. Using magnetic resonance chemical shift coding imaging technology, magnetic resonance gradient echo imaging sequences are used to collect multiple magnetic resonance images with different echo times. The water hydrogen protons and aliphatic hydrogen protons have different phase differences in the magnetic resonance images under time. Then, by solving the preset coding imaging model, water hydrogen protons and aliphatic hydrogen protons can be separated to obtain pure water images and pure fat images, and then protons can be obtained according to the proton density fat fraction = fat image/(fat image + water image) Density fat fraction.

在实际应用中，终端设备可以与磁共振扫描仪通信连接，磁共振扫描仪在采集到目标图像后，能够将采集到的目标图像发送给终端设备。具体地，可以通过设置磁共振扫描仪的磁共振扫描参数，然后控制磁共振扫描仪连续扫描被测对象的预设部位(覆盖全身)，使得磁共振扫描仪采集到各个预设部位的多个不同回波时间的磁共振图像，最终输出被测对象的全身脂肪分布的质子密度脂肪分数定量图，即目标图像。In practical applications, the terminal device can be connected to the magnetic resonance scanner in communication, and after the magnetic resonance scanner acquires the target image, the acquired target image can be sent to the terminal device. Specifically, the magnetic resonance scanning parameters of the magnetic resonance scanner can be set, and then the magnetic resonance scanner can be controlled to continuously scan the preset parts (covering the whole body) of the measured object, so that the magnetic resonance scanner can collect multiple preset parts Magnetic resonance images of different echo times, and finally output the proton density fat fraction quantitative map of the whole body fat distribution of the measured object, that is, the target image.

在本申请实施例的一种可能的实现方式中，可以将上述磁共振扫描参数设置为三维磁共振FLASH序列，进行六回波数据采集，重复时间为10.5ms，回波时间分别设置为1.67ms/3.15ms/4.63ms/6.11ms/7.59ms/9.07ms，并将翻转角为3°，视野设置为300mm×400mm，图像矩阵大小126×224，层厚6mm，每个部位采集20层数据。In a possible implementation manner of the embodiment of the present application, the above-mentioned magnetic resonance scanning parameters may be set as a three-dimensional magnetic resonance FLASH sequence, and six-echo data acquisition is performed, the repetition time is 10.5ms, and the echo time is respectively set to 1.67ms /3.15ms/4.63ms/6.11ms/7.59ms/9.07ms, the flip angle was set to 3°, the field of view was set to 300mm×400mm, the image matrix size was 126×224, the slice thickness was 6mm, and 20 slices of data were collected for each part.

在本申请实施例中，上述被测对象的预设部位可以是覆盖被测对象的颈部到膝盖的部位。可以理解的是，上述被测对象的预设部位还可以根据测试的需求进行设置，在此不加以限制。In the embodiment of the present application, the preset part of the measured object may be a part covering the neck to the knee of the measured object. It can be understood that, the preset part of the object to be tested can also be set according to the requirements of the test, which is not limited here.

具体的，上述预设的编码成像模型如下：Specifically, the above-mentioned preset coding imaging model is as follows:

其中，TE _n是指回波时间，S _n是在回波时间TE _n下的信号强度，N为回波个数，ρ _W是指水的信号强度值，ρ _f是指脂肪的信号强度值；P为脂肪的波峰分量的个数，a _P是指每个波峰分量对应的相对幅度，f _F,P是水中氢质子与脂肪氢质子的化学位移差，f _B为局部主磁场不均匀参数。 Among them, TE _n is the echo time, _Sn is the signal strength at the echo time TE _n , N is the number of echoes, ρ _W is the signal strength value of water, ρ _f is the signal strength value of fat ; P is the number of peak components of fat, a _P refers to the relative amplitude corresponding to each peak component, f _{F, P} is the chemical shift difference between hydrogen protons in water and fat hydrogen protons, f _B is the local main magnetic field inhomogeneity parameter .

在本申请实施例中，回波个数N大于等于6；每个波峰分量对应的相对幅度a _P满足

水中氢质子与脂肪氢质子的化学位移差f _F,P与温度成正比，例如在人体37摄氏度的环境下，二者之间的化学位移差是-3.35ppm；在仿体20摄氏度的温度下，二者之间的化学位移差是-3.52ppm；局部主磁场不均匀参数f _B＝γΔB ₀，γ是氢质子旋磁比，γ＝42.576MHz/T；B ₀是磁共振的主磁场强度，ΔB ₀是由于***误差和被测对象的影响等因素，导致主磁场不完全均匀，而存在的磁场内部的局部变化量。 In the embodiment of the present application, the number N of echoes is greater than or equal to 6; the relative amplitude a _P corresponding to each peak component satisfies

The chemical shift difference f _{F, P} of hydrogen protons in water and aliphatic hydrogen protons is proportional to temperature. For example, in the environment of 37 degrees Celsius of the human body, the chemical shift difference between the two is -3.35ppm; , the chemical shift difference between the two is -3.52ppm; the local main magnetic field inhomogeneous parameter f _B =γΔB ₀ , γ is the hydrogen proton gyromagnetic ratio, γ = 42.576MHz/T; B ₀ is the main magnetic field strength of the magnetic resonance , ΔB ₀ is the local variation within the existing magnetic field due to the fact that the main magnetic field is not completely uniform due to factors such as systematic errors and the influence of the measured object.

在本申请实施例中，通过获取多个不同回波时间对应的磁共振图像，就能够得到确定出每张磁共振图像中的水中氢质子与脂肪氢质子的化学位移差f _F,P，脂肪的波峰分量的相对幅度a _P，回波时间TE _n和回波时间TE _n下的信号强度S _n，通过上述预设的编码成像模型就能够求解得到水的信号强度值ρ _W，脂肪的信号强度值ρ _f，磁场内部的局部变化量ΔB ₀。 In the embodiment of the present application, by acquiring a plurality of magnetic resonance images corresponding to different echo times, the chemical shift difference f _F,P between the hydrogen protons in water and the aliphatic hydrogen protons in each magnetic resonance image can be determined, and the fat The relative amplitude a _P of the wave peak component, the echo time TE _n and the signal strength S _n at the echo time TE _n can be solved by the above preset coding imaging model to obtain the signal strength value ρ _W of water, the signal of fat The intensity value ρ _f , the local variation ΔB ₀ inside the magnetic field.

通过确定出水的信号强度值ρ _W和脂肪的信号强度值ρ _f之后，基于质子密度脂肪分数＝脂肪的信号强度/(脂肪的信号强度+水的信号强度)确定出质子密度脂肪分数。能够精准地确定出全身脂肪组织的体积占比，实现了全身脂肪组织的准确定量。 After determining the signal strength value _ρW of water and the signal strength value _ρf of fat, the proton density fat fraction is determined based on proton density fat fraction=fat signal strength/(fat signal strength+water signal strength). It can accurately determine the volume ratio of the whole body adipose tissue, and realize the accurate quantification of the whole body adipose tissue.

S12：将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。S12: Input the target image into a preset image segmentation model for processing, and segment the subcutaneous fat image and the visceral fat image of each preset part from the target image.

本申请实施例中，在获取到能够准确定量全身脂肪组织占比的目标图像后，通过图像分割模型对目标图像进行分割。就准确地能够得到全身脂肪的分布情况，确定出各个脏器组织的脂肪沉积情况。In the embodiment of the present application, after obtaining a target image that can accurately quantify the proportion of whole body adipose tissue, the target image is segmented by an image segmentation model. The distribution of body fat can be accurately obtained, and the fat deposition of each organ tissue can be determined.

上述图像分割模型用于将目标图像中被测对象的各个预设部位的皮下脂肪图像和内脏脂肪图像进行分割，即图像分割模型的输入为目标图像，输出为目标图像中各个预设部位的皮下脂肪图像和内脏脂肪图像。The above image segmentation model is used to segment the subcutaneous fat image and visceral fat image of each preset part of the measured object in the target image, that is, the input of the image segmentation model is the target image, and the output is the subcutaneous fat image of each preset part in the target image. Fat image and visceral fat image.

请参阅图2，图2示出了本申请实施例提供的图像分割方法的实现过程。在本申请实施例中，将目标图像输入到图像分割模型后，图像分割模型可以从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图，然后基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图(包括皮下脂肪和内脏脂肪)，根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，最后输出各个预设部位对应的皮下脂肪图和内脏脂肪图(主要是腹部内脏脂肪图)。Referring to FIG. 2, FIG. 2 shows an implementation process of the image segmentation method provided by the embodiment of the present application. In the embodiment of the present application, after the target image is input into the image segmentation model, the image segmentation model can identify the accurate quantitative map of proton density and fat fraction corresponding to each preset part from the target image, and then based on the accurate proton density and fat fraction corresponding to each preset part The proton density fat fraction quantitative map determines the overall fat map (including subcutaneous fat and visceral fat) corresponding to each preset part, and segments the subcutaneous fat map and visceral fat corresponding to each preset part according to the total fat map corresponding to each preset part Figure, and finally output the subcutaneous fat map and visceral fat map (mainly abdominal visceral fat map) corresponding to each preset part.

在本申请实施例中，根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，具体可以包括以下步骤：In the embodiment of the present application, the subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part, which may specifically include the following steps:

对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；再对边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到分割结果图(即各个预设部位对应的皮下脂肪图和内脏脂肪图)。Perform down-sampling operation on the overall fat map corresponding to each preset part, extract edge features of the overall fat map corresponding to each preset part, and obtain an edge feature map; and then perform an up-sampling operation on the edge feature map to restore all the edge features. The spatial information and edge information in the overall fat map corresponding to each preset part are described to obtain a segmentation result map (ie, the subcutaneous fat map and the visceral fat map corresponding to each preset part).

请参阅图3，图3是本申请实施例提供的一种图像分割模型的结构示意图。如图3所示，在本申请的一个实施例中，图像分割模型30可以包括下采样网络31和上采样网络32。Please refer to FIG. 3 , which is a schematic structural diagram of an image segmentation model provided by an embodiment of the present application. As shown in FIG. 3 , in one embodiment of the present application, the image segmentation model 30 may include a downsampling network 31 and an upsampling network 32 .

下采样网络31用于提取各个预设部位对应的皮下脂肪图和内脏脂肪图的边缘特征。所述边缘特征能够用来表明皮下脂肪和内脏脂肪的边界。The downsampling network 31 is used to extract edge features of the subcutaneous fat map and the visceral fat map corresponding to each preset part. The edge features can be used to indicate the boundaries of subcutaneous fat and visceral fat.

上采样网络32用于恢复各个预设部位对应的皮下脂肪图和内脏脂肪图中的空间信息和边缘信息。The upsampling network 32 is used to restore the spatial information and edge information in the subcutaneous fat map and the visceral fat map corresponding to each preset part.

在实际应用中，图像分割模型可以使用跳跃连接网络(skip connection)连接下采样网络31与上采样网络32，使得下采样网络31提取的特征可以直接传递到上采样网络32，解决训练的过程中梯度消失的问题。In practical applications, the image segmentation model can use a skip connection to connect the downsampling network 31 and the upsampling network 32, so that the features extracted by the downsampling network 31 can be directly transferred to the upsampling network 32. The problem of vanishing gradients.

本申请实施例中的图像分割模型可以是基于预设的样本数据集，采用深度学习的方式对预先构建的图像分割模型进行训练得到的。The image segmentation model in the embodiment of the present application may be obtained by training a pre-built image segmentation model by using a deep learning method based on a preset sample data set.

在本申请另一实施例中，上述图像分割方法还可以包括以下步骤：In another embodiment of the present application, the above-mentioned image segmentation method may further include the following steps:

构建图像分割模型；Build an image segmentation model;

获取样本数据集；Get a sample dataset;

本申请实施例中，可以构建网络结构如图3所示的图像分割模型，图像分割模型中涉及到的每个网络参数(例如卷积核的各项参数)的初始值可以是随机赋予的任意值，图像分割模型涉及到的每个网络参数的最终值可以是在对图像分割模型的训练过程中学习得到的。In this embodiment of the present application, an image segmentation model with a network structure as shown in FIG. 3 can be constructed, and the initial value of each network parameter (for example, parameters of the convolution kernel) involved in the image segmentation model can be any random assigned value. The final value of each network parameter involved in the image segmentation model can be learned during the training process of the image segmentation model.

请参阅图4，上述图像分割模型中，下采样网络31可以包括第一下采样层、第二下采样层、第三下采样层和第四下采样层，上采样网络32可以包括第一上采样层、第二上采样层、第三上采样层和第四上采样层。Referring to FIG. 4, in the above image segmentation model, the downsampling network 31 may include a first downsampling layer, a second downsampling layer, a third downsampling layer and a fourth downsampling layer, and the upsampling network 32 may include a first upsampling layer A sampling layer, a second upsampling layer, a third upsampling layer, and a fourth upsampling layer.

请继续参阅图4，为了解决梯度消失的问题，上述图像分割模型还可以包括连接在第一下采样层与第四上采样层之间的第一跳跃连接网络、连接在第二下采样层与第三上采样层之间的第二跳跃连接网络、连接在第三下采样层与第二上采样层之间的第三跳跃连接网络以及连接在第四下采样层与第一上采样层之间的第四跳跃结构。Please continue to refer to FIG. 4. In order to solve the problem of gradient disappearance, the above-mentioned image segmentation model may further include a first skip connection network connected between the first downsampling layer and the fourth upsampling layer, a first skip connection network connected between the second downsampling layer and the fourth upsampling layer. a second skip connection network between the third upsampling layer, a third skip connection network connected between the third downsampling layer and the second upsampling layer, and a third skip connection network connected between the fourth downsampling layer and the first upsampling layer The fourth jump structure between.

在具体应用中，上述第一下采样层可以包括两个3*3卷积层和一个2*2的最大池化层；第二下采样层同样包括两个3*3卷积层和一个2*2的最大池化层；第三下采样层也包括两个3*3卷积层和一个2*2的最大池化层；第四下采样层也包括两个3*3卷积层和一个2*2的最大池化层；第一上采样层可以包括两个3*3卷积层和一个2*2上采样层；第二上采样层同样包括两个3*3卷积层和一个2*2上采样层，第三上采样层也包括两个3*3卷积层和一个2*2上采样层；第四上采样层也包括两个3*3卷积层和一个2*2上采样层。In a specific application, the above-mentioned first downsampling layer may include two 3*3 convolutional layers and a 2*2 maximum pooling layer; the second downsampling layer also includes two 3*3 convolutional layers and a 2*2 convolutional layer. *2 max pooling layer; the third downsampling layer also includes two 3*3 convolutional layers and a 2*2 max pooling layer; the fourth downsampling layer also includes two 3*3 convolutional layers and A 2*2 max pooling layer; the first upsampling layer may include two 3*3 convolutional layers and a 2*2 upsampling layer; the second upsampling layer also includes two 3*3 convolutional layers and A 2*2 upsampling layer, the third upsampling layer also includes two 3*3 convolutional layers and a 2*2 upsampling layer; the fourth upsampling layer also includes two 3*3 convolutional layers and a 2 *2 Upsampling layer.

在实际应用中，上述跳跃连接网络通过将各个下采样层对应位置的特征图复制并剪切到上采样的过程中，使得底层特征与高层特征得以融合，以保留更多的高分辨率细节信息，提高图像分割精度。In practical applications, the above skip connection network copies and cuts the feature maps of the corresponding positions of each down-sampling layer into the up-sampling process, so that the low-level features and high-level features are fused to retain more high-resolution details. , to improve the image segmentation accuracy.

对应地，上述对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图具体可以是：通过第一下采样层对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；通过第二下采样层对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；通过第三下采样层对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；通过第四下采样层对所述第三下采样特征图进行下采样处理，得到边缘特征图。Correspondingly, the above-mentioned down-sampling operation is performed on the overall fat map corresponding to each preset part, and edge features of the overall fat map corresponding to each preset part are extracted to obtain the edge feature map. Specifically, the first downsampling may be performed. The second downsampling layer performs downsampling processing on the overall fat map corresponding to each preset part to obtain a first downsampling feature map; the second downsampling layer performs downsampling processing on the first downsampling feature map to obtain a second downsampling feature map. Sampling a feature map; performing down-sampling processing on the second down-sampling feature map through a third down-sampling layer to obtain a third down-sampling feature map; performing down-sampling on the third down-sampling feature map through a fourth down-sampling layer process to obtain the edge feature map.

对应地，上述对边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到分割结果图具体可以是：通过第一上采样层对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第四跳跃连接网络复制的第一通道特征图进行融合，得到第一融合特征图；通过第二上采样层对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第三跳跃连接网络复制的第二通道特征图进行融合，得到第二融合特征图；通过第三上采样层对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第二跳跃连接网络复制的第三通道特征图进行融合，得到第三融合特征图；通过第四上采样层对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第一跳跃连接网络复制的第四通道特征图进行融合，得到分割结果图。Correspondingly, the above-mentioned upsampling operation is performed on the edge feature map, the spatial information and edge information in the overall fat map corresponding to each preset part are restored, and the obtained segmentation result map may specifically be: The first upsampling feature map is obtained, and the first upsampling feature map is fused with the first channel feature map copied by the fourth skip connection network to obtain the first fusion feature map; through the second upsampling The layer performs up-sampling processing on the first fusion feature map to obtain a second up-sampling feature map, and fuses the second sampling feature map with the second channel feature map copied by the third skip connection network to obtain a second fusion feature. Figure; the second fusion feature map is subjected to upsampling processing by the third upsampling layer to obtain the third upsampling feature map, and the third upsampling feature map and the third channel feature map copied by the second skip connection network Perform fusion to obtain a third fusion feature map; perform upsampling processing on the third fusion feature map through the fourth upsampling layer to obtain a fourth upsampling feature map, and copy the fourth upsampling feature map with the first skip connection network The fourth channel feature map is fused to obtain the segmentation result map.

需要说明的是，上述第四通道特征图是基于第一下采样层的卷积层对各个预设部位对应的总体脂肪图进行卷积处理后得到的通道特征图，上述第三通道特征图是基于第二下采样层的卷积层对第一下采样特征图进行卷积处理后得到的通道特征图；上述第二通道特征图是基于第三下采样层的卷积层对第二下采样特征图进行卷积处理后得到的通道特征图；上述第一通道特征图是基于第四下采样层的卷积层对第三下采样特征图进行卷积处理后得到的通道特征图。It should be noted that the above-mentioned fourth channel feature map is a channel feature map obtained by convolving the overall fat map corresponding to each preset part based on the convolution layer of the first downsampling layer, and the above-mentioned third channel feature map is: The channel feature map obtained by convolution of the first downsampling feature map based on the convolutional layer of the second downsampling layer; the second channel feature map is based on the convolutional layer of the third downsampling layer on the second downsampling layer The channel feature map obtained after the feature map is convolved; the above-mentioned first channel feature map is a channel feature map obtained by convolution of the third down-sampling feature map based on the convolution layer of the fourth down-sampling layer.

在构建了图像分割模型后，可以通过网络上海量的医疗图像资源来获取样本数据集。After the image segmentation model is constructed, the sample data set can be obtained through a large amount of medical image resources on the Internet.

本申请实施例中，以覆盖人体颈部到膝盖部位的质子密度脂肪分数定量图(样本图像) 和对应于质子密度脂肪分数定量图的皮下脂肪图和内脏脂肪图(标签图像)作为一组样本数据。In the embodiment of the present application, the proton density fat fraction quantitative map (sample image) covering the human body from the neck to the knee, and the subcutaneous fat map and visceral fat map (label image) corresponding to the proton density fat fraction quantitative map are used as a set of samples data.

在具体应用中，可以基于专业人士手动在覆盖人体颈部到膝盖部位的质子密度脂肪分数定量图中分别圈画出各个预设部位的皮下脂肪图像和内脏脂肪内脏作为质子密度脂肪分数定量图对应的标签图像。In specific applications, the subcutaneous fat image and visceral fat viscera of each preset part can be circled and drawn in the proton density fat fraction quantitative map covering the human neck to the knees manually based on professionals as the corresponding proton density fat fraction quantitative map. label image.

在实际应用中，可以选用不小于1000组的样本数据，得到样本数据集。将样本数据集分为训练集、验证集和测试集。为了达到训练要求，可以将50％的样本数据作为训练集，剩余部分作为验证集和测试集。In practical applications, no less than 1000 groups of sample data can be selected to obtain a sample data set. Divide the sample dataset into training set, validation set and test set. In order to meet the training requirements, 50% of the sample data can be used as the training set, and the rest as the validation set and test set.

在获取到样本数据后，通过训练集数据对该图像分割模型进行训练，并使用验证集进行快速调参，再使用测试集对图像分割模型进行测试，得到训练完成的图像分割模型。After obtaining the sample data, the image segmentation model is trained through the training set data, and the validation set is used to quickly adjust the parameters, and then the test set is used to test the image segmentation model, and the trained image segmentation model is obtained.

在训练图像分割模型时，可以将样本图像输入预先构建的图像分割模型进行处理，得到该样本图像对应的分割结果图。然后再基于样本数据的标签图像与图像分割模型输出的分割结果图对上述图像分割模型中的网络参数进行调整，以得到能够使得该图像分割模型的loss函数收敛网络参数。再基于验证集和测试集中的样本数据对调整完网络参数的图像分割模型进行验证和测试，验证和测试通过即说明图像分割模型训练完成。When training an image segmentation model, a sample image can be input into a pre-built image segmentation model for processing, and a segmentation result map corresponding to the sample image can be obtained. Then, based on the label image of the sample data and the segmentation result map output by the image segmentation model, the network parameters in the above image segmentation model are adjusted to obtain network parameters that can make the loss function of the image segmentation model converge. Then, based on the sample data in the verification set and the test set, the image segmentation model after adjusting the network parameters is verified and tested. Passing the verification and test means that the training of the image segmentation model is completed.

终端设备可以将训练完成的图像分割模型确定为预设的图像分割模型，也即S12中所述的预设的图像分割模型。The terminal device may determine the trained image segmentation model as a preset image segmentation model, that is, the preset image segmentation model described in S12.

以上可以看出，本申请实施例提供的图像分割方法，通过获取能够精准定量的全身脂肪分布的质子密度脂肪分数定量图，再通过预设的图像分割模型对全身脂肪分布的质子密度脂肪分数定量图进行处理，进而实现对人体的各个部分的脂肪进行准确分割。解决了现有的脂肪定量和分割方法通常是针对某一部位进行处理，无法实现全身脂肪的分割和全身脂肪沉积程度的定量和全身脂肪的准确分割的问题。It can be seen from the above that the image segmentation method provided by the embodiments of the present application obtains a quantitative map of proton density fat fraction of whole body fat distribution that can be accurately quantified, and then uses a preset image segmentation model to quantify the proton density fat fraction of whole body fat distribution. The image is processed, and then the fat of each part of the human body can be accurately segmented. It solves the problem that the existing fat quantification and segmentation methods usually deal with a certain part, and cannot achieve the segmentation of whole body fat, the quantification of the degree of whole body fat deposition and the accurate segmentation of whole body fat.

应理解，上述实施例中各步骤的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请实施例的实施过程构成任何限定。It should be understood that the size of the sequence numbers of the steps in the above embodiments does not mean the sequence of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

基于上述实施例所提供的图像分割方法，本发明实施例进一步给出实现上述方法实施例的终端设备的实施例。Based on the image segmentation method provided by the foregoing embodiment, the embodiment of the present invention further provides an embodiment of a terminal device implementing the foregoing method embodiment.

请参阅图5，图5是本申请实施例提供的一种终端设备的结构示意图。本申请实施例中，终端设备包括的各单元用于执行图1至图4对应的实施例中的各步骤。具体请参阅图1至图4以及图1至图4对应的实施例中的相关描述。为了便于说明，仅示出了与本实施例相关的部分。如图5所示，终端设备50包括：第一获取单元51和第一处理单元52。其中：Please refer to FIG. 5. FIG. 5 is a schematic structural diagram of a terminal device provided by an embodiment of the present application. In this embodiment of the present application, each unit included in the terminal device is used to perform each step in the embodiment corresponding to FIG. 1 to FIG. 4 . For details, please refer to FIG. 1 to FIG. 4 and the related descriptions in the corresponding embodiments of FIG. 1 to FIG. 4 . For convenience of explanation, only the parts related to this embodiment are shown. As shown in FIG. 5 , the terminal device 50 includes: a first obtaining unit 51 and a first processing unit 52 . in:

第一获取单元51用于获取目标图像。所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图。The first acquisition unit 51 is used to acquire the target image. The target image is the proton density fat fraction quantitative map of the whole body fat distribution of the measured object.

第一处理单元52用于将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The first processing unit 52 is configured to input the target image into a preset image segmentation model for processing, and segment the subcutaneous fat image and the visceral fat image of each preset part from the target image.

可选的，所述第一获取单元包括第二获取单元和第一确定单元。Optionally, the first obtaining unit includes a second obtaining unit and a first determining unit.

第二获取单元用于获取多个不同回波时间的磁共振图像。The second acquisition unit is used for acquiring a plurality of magnetic resonance images with different echo times.

第一确定单元用于根据所述多个不同回波时间的磁共振图像确定目标图像。The first determination unit is configured to determine a target image according to the plurality of magnetic resonance images with different echo times.

可选的，上述第一处理单元52包括识别单元、第二确定单元和分割单元。Optionally, the above-mentioned first processing unit 52 includes an identification unit, a second determination unit and a segmentation unit.

识别单元用于从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；The identification unit is used to identify the accurate proton density fat fraction quantitative map corresponding to each preset part from the target image;

第二确定单元用于基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；The second determination unit is configured to determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

分割单元用于根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The segmentation unit is configured to segment the subcutaneous fat map and the visceral fat map corresponding to each preset part according to the overall fat map corresponding to each preset part.

可选的，上述分割单元可以包括下采样单元和上采样单元。Optionally, the above-mentioned dividing unit may include a down-sampling unit and an up-sampling unit.

上述下采样单元用于对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；The above-mentioned down-sampling unit is configured to perform a down-sampling operation on the overall fat map corresponding to each preset part, and extract edge features of the overall fat map corresponding to each preset part to obtain an edge feature map;

上述上采样单元用于对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图。The above-mentioned up-sampling unit is used to perform an up-sampling operation on the edge feature map, recover the spatial information and edge information in the overall fat map corresponding to each preset part, and obtain the subcutaneous fat map and visceral fat map corresponding to each preset part picture.

可选地，上述下采样单元具体用于对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；对所述第三下采样特征图进行下采样处理，得到边缘特征图。Optionally, the above-mentioned down-sampling unit is specifically configured to perform down-sampling processing on the overall fat map corresponding to each preset part to obtain a first down-sampling feature map; and perform down-sampling processing on the first down-sampling feature map, obtaining a second down-sampling feature map; performing down-sampling processing on the second down-sampling feature map to obtain a third down-sampling feature map; performing down-sampling processing on the third down-sampling feature map to obtain an edge feature map.

可选地，上述上采样单元具体用于对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第一通道特征图进行融合，得到第一融合特征图；对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第二通道特征图进行融合，得到第二融合特征图；对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第三通道特征图进行融合，得到第三融合特征图；对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第四通道特征图进行融合，得到分割结果图。Optionally, the above-mentioned upsampling unit is specifically configured to perform upsampling processing on the edge feature map to obtain the first upsampling feature map, and fuse the first upsampling feature map with the first channel feature map to obtain the first fusion feature. Figure; perform up-sampling processing on the first fusion feature map to obtain a second up-sampling feature map, and fuse the second sampling feature map with the second channel feature map to obtain a second fusion feature map; Perform upsampling processing on the second fusion feature map to obtain a third upsampling feature map, and fuse the third upsampling feature map with the third channel feature map to obtain a third fusion feature map; perform upsampling on the third fusion feature map processing to obtain a fourth upsampling feature map, and fusing the fourth upsampling feature map with the fourth channel feature map to obtain a segmentation result map.

可选的，所述终端设备还包括构建单元、第三获取单元、训练单元。Optionally, the terminal device further includes a construction unit, a third acquisition unit, and a training unit.

构建单元用于构建图像分割模型。The building unit is used to build an image segmentation model.

第三获取单元用于获取样本数据集；其中，所述样本数据集包括多组样本数据，每组样本数据包括样本图像和标签图像，所述样本图像是指覆盖人体颈部到膝盖部位的质子密度脂肪分数定量图，所述标签图像是指对应于质子密度脂肪分数定量图的皮下脂肪图和内脏脂肪图。The third acquiring unit is used to acquire a sample data set; wherein, the sample data set includes multiple groups of sample data, each group of sample data includes a sample image and a label image, and the sample image refers to the protons covering the parts from the neck to the knee of the human body Density fat fraction quantification map, the label images refer to subcutaneous fat map and visceral fat map corresponding to the proton density fat fraction quantification map.

训练单元用于基于样本数据集对所述预先构建的图像分割模型进行训练，得到所述预设图像分割模型。The training unit is configured to train the pre-built image segmentation model based on the sample data set to obtain the preset image segmentation model.

需要说明的是，上述模块之间的信息交互、执行过程等内容，由于与本申请方法实施例基于同一构思，其具体功能及带来的技术效果，具体可参照方法实施例部分，此处不再赘述。It should be noted that the information exchange, execution process and other contents between the above modules are based on the same concept as the method embodiments of the present application. For specific functions and technical effects, please refer to the method embodiments section for details. Repeat.

图6是本申请另一实施例提供的一种终端设备的结构示意图。如图6所示，该实施例提供的终端设备6包括：处理器60、存储器61以及存储在所述存储器61中并可在所述处理器60上运行的计算机程序62，例如图像分割程序。处理器60执行所述计算机程序62时实现上述各个图像分割方法实施例中的步骤，例如图1所示的S11～S12。或者，所述处理器60执行所述计算机程序62时实现上述各终端设备实施例中各模块/单元的功能，例如图5所示单元51～52的功能。FIG. 6 is a schematic structural diagram of a terminal device provided by another embodiment of the present application. As shown in FIG. 6 , the terminal device 6 provided in this embodiment includes: a processor 60 , a memory 61 , and a computer program 62 stored in the memory 61 and executable on the processor 60 , such as an image segmentation program. When the processor 60 executes the computer program 62, the steps in each of the foregoing image segmentation method embodiments are implemented, for example, S11 to S12 shown in FIG. 1 . Alternatively, when the processor 60 executes the computer program 62, the functions of the modules/units in each of the foregoing terminal device embodiments, such as the functions of the units 51 to 52 shown in FIG. 5, are implemented.

示例性的，所述计算机程序62可以被分割成一个或多个模块/单元，所述一个或者多个模块/单元被存储在所述存储器61中，并由处理器60执行，以完成本申请。所述一个或多个模块/单元可以是能够完成特定功能的一系列计算机程序指令段，该指令段用于描述所述计算机程序62在所述终端设备6中的执行过程。例如，所述计算机程序62可以被分割成第一获取单元和第一处理单元，各单元具体功能请参阅图5对应地实施例中的相关描述，此处不赘述。Exemplarily, the computer program 62 may be divided into one or more modules/units, and the one or more modules/units are stored in the memory 61 and executed by the processor 60 to complete the present application. . The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 62 in the terminal device 6 . For example, the computer program 62 may be divided into a first obtaining unit and a first processing unit, and the specific functions of each unit can be referred to the relevant description in the corresponding embodiment in FIG. 5 , and details are not repeated here.

所述终端设备可包括但不仅限于，处理器60、存储器61。本领域技术人员可以理解，图6仅仅是终端设备6的示例，并不构成对终端设备6的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件，例如所述终端设备还可以包括输入输出设备、网络接入设备、总线等。The terminal device may include, but is not limited to, the processor 60 and the memory 61 . Those skilled in the art can understand that FIG. 6 is only an example of the terminal device 6, and does not constitute a limitation on the terminal device 6, and may include more or less components than the one shown, or combine some components, or different components For example, the terminal device may further include an input and output device, a network access device, a bus, and the like.

所称处理器60可以是中央处理单元(Central Processing Unit，CPU)，还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现成可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。The so-called processor 60 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

所述存储器61可以是所述终端设备6的内部存储单元，例如终端设备6的硬盘或内存。所述存储器61也可以是所述终端设备6的外部存储设备，例如所述终端设备6上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。进一步地，所述存储器61还可以既包括所述终端设备6的内部存储单元也包括外部存储设备。所述存储器61用于存储所述计算机程序以及所述终端设备所需的其他程序和数据。所述存储器61还可以用于暂时地存储已经输出或者将要输出的数据。The memory 61 may be an internal storage unit of the terminal device 6 , such as a hard disk or a memory of the terminal device 6 . The memory 61 can also be an external storage device of the terminal device 6, such as a plug-in hard disk equipped on the terminal device 6, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, Flash Card, etc. Further, the memory 61 may also include both an internal storage unit of the terminal device 6 and an external storage device. The memory 61 is used to store the computer program and other programs and data required by the terminal device. The memory 61 can also be used to temporarily store data that has been output or will be output.

本申请实施例还提供了一种计算机可读存储介质。请参阅图7，图7是本申请实施例提供的一种计算机可读存储介质的结构示意图，如图7所示，计算机可读存储介质70中存储有计算机程序71，计算机程序71被处理器执行时可实现上述图像分割方法。Embodiments of the present application also provide a computer-readable storage medium. Please refer to FIG. 7. FIG. 7 is a schematic structural diagram of a computer-readable storage medium provided by an embodiment of the present application. As shown in FIG. 7, a computer program 71 is stored in the computer-readable storage medium 70, and the computer program 71 is processed by a processor. The above-mentioned image segmentation method can be realized when executed.

本申请实施例提供了一种计算机程序产品，当计算机程序产品在终端设备上运行时，使得终端设备执行时实现可实现上述图像分割方法。The embodiments of the present application provide a computer program product, when the computer program product runs on a terminal device, the above-mentioned image segmentation method can be implemented when the terminal device executes.

所属领域的技术人员可以清楚地了解到，为了描述的方便和简洁，仅以上述各功能单元、模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能单元、模块完成，即将所述终端设备的内部结构划分成不同的功能单元或模块，以完成以上描述的全部或者部分功能。实施例中的各功能单元、模块可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中，上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。另外，各功能单元、模块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。上述***中单元、模块的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. The module is completed, that is, the internal structure of the terminal device is divided into different functional units or modules, so as to complete all or part of the functions described above. Each functional unit and module in the embodiment may be integrated in one processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit, and the above-mentioned integrated units may adopt hardware. It can also be realized in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing from each other, and are not used to limit the protection scope of the present application. For the specific working process of the units and modules in the above-mentioned system, reference may be made to the corresponding process in the foregoing method embodiments, which will not be repeated here.

在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述或记载的部分，可以参照其它实施例的相关描述。In the foregoing embodiments, the description of each embodiment has its own emphasis. For parts that are not described or described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each particular application, but such implementations should not be considered beyond the scope of this application.

以上所述实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围，均应包含在本申请的保护范围之内。The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the above-mentioned embodiments, those of ordinary skill in the art should understand that: it can still be used for the above-mentioned implementations. The technical solutions described in the examples are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the application, and should be included in the within the scope of protection of this application.

Claims

一种图像分割方法，其特征在于，包括：An image segmentation method, comprising:

获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；Acquiring a target image; the target image is a proton density fat fraction quantitative map of the body fat distribution of the subject;

将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The target image is input into a preset image segmentation model for processing, and subcutaneous fat images and visceral fat images of each preset part are segmented from the target image.
根据权利要求1所述的图像分割方法，其特征在于，所述获取目标图像，包括：The image segmentation method according to claim 1, wherein the acquiring the target image comprises:

获取多个不同回波时间的磁共振图像；Acquiring multiple magnetic resonance images with different echo times;

根据所述多个不同回波时间的磁共振图像确定目标图像。The target image is determined from the plurality of magnetic resonance images with different echo times.
根据权利要求1所述的图像分割方法，其特征在于，所述将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图，包括：The image segmentation method according to claim 1, wherein the target image is input into a preset image segmentation model for processing, and subcutaneous fat images of each preset part are segmented from the target image and visceral fat maps, including:

从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；Identify the precise proton density fat fraction quantitative map corresponding to each preset part from the target image;

基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；Determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part.
根据权利要求3所述的图像分割方法，其特征在于，所述根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The image segmentation method according to claim 3, wherein the segmenting the subcutaneous fat map and the visceral fat map corresponding to each preset part according to the overall fat map corresponding to each preset part comprises:

对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；performing a downsampling operation on the overall fat map corresponding to each preset part, and extracting edge features of the total fat map corresponding to each preset part to obtain an edge feature map;

对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图。An up-sampling operation is performed on the edge feature map, the spatial information and edge information in the overall fat map corresponding to each preset part are restored, and a subcutaneous fat map and a visceral fat map corresponding to each preset part are obtained.
根据权利要求4所述的图像分割方法，其特征在于，所述对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图，包括：The image segmentation method according to claim 4, wherein the downsampling operation is performed on the overall fat map corresponding to each preset part, and the edge features of the total fat map corresponding to each preset part are extracted, Get the edge feature map, including:

对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；Perform down-sampling processing on the overall fat map corresponding to each preset part to obtain a first down-sampling feature map;

对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；performing down-sampling processing on the first down-sampling feature map to obtain a second down-sampling feature map;

对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；performing down-sampling processing on the second down-sampling feature map to obtain a third down-sampling feature map;

对所述第三下采样特征图进行下采样处理，得到边缘特征图。Perform down-sampling processing on the third down-sampling feature map to obtain an edge feature map.
根据权利要求5所述的图像分割方法，其特征在于，所述对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The image segmentation method according to claim 5, wherein the upsampling operation is performed on the edge feature map to restore the spatial information and edge information in the overall fat map corresponding to each preset part, and obtain each Subcutaneous fat map and visceral fat map corresponding to preset parts, including:

对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第一通道特征图进行融合，得到第一融合特征图；Perform up-sampling processing on the edge feature map to obtain a first up-sampling feature map, and fuse the first up-sampling feature map with the first channel feature map to obtain a first fused feature map;

对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第二通道特征图进行融合，得到第二融合特征图；Perform up-sampling processing on the first fusion feature map to obtain a second up-sampling feature map, and fuse the second sampling feature map with the second channel feature map to obtain a second fusion feature map;

对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第三通道特征图进行融合，得到第三融合特征图；Perform upsampling processing on the second fusion feature map to obtain a third upsampling feature map, and fuse the third upsampling feature map with the third channel feature map to obtain a third fusion feature map;

对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第四通道特征图进行融合，得到分割结果图。Perform up-sampling processing on the third fusion feature map to obtain a fourth up-sampling feature map, and fuse the fourth up-sampling feature map with the fourth channel feature map to obtain a segmentation result map.
根据权利要求1至6任意一项所述的图像分割方法，其特征在于，还包括：The image segmentation method according to any one of claims 1 to 6, further comprising:

构建图像分割模型；Build an image segmentation model;

获取样本数据集；其中，所述样本数据集包括多组样本数据，每组样本数据包括样本图像和标签图像，所述样本图像是指覆盖人体颈部到膝盖部位的质子密度脂肪分数定量图，所述标签图像是指对应于质子密度脂肪分数定量图的皮下脂肪图和内脏脂肪图；Obtaining a sample data set; wherein, the sample data set includes multiple sets of sample data, each set of sample data includes a sample image and a label image, and the sample image refers to a quantitative map of proton density fat fraction covering the part of the human body from the neck to the knee, The label image refers to a subcutaneous fat map and a visceral fat map corresponding to a quantitative map of proton density fat fraction;

基于样本数据集对所述构建的图像分割模型进行训练，得到所述预设图像分割模型。The constructed image segmentation model is trained based on the sample data set to obtain the preset image segmentation model.
一种终端设备，其特征在于，包括：A terminal device, characterized in that it includes:

第一获取单元，用于获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；The first acquisition unit is used to acquire a target image; the target image is a proton density fat fraction quantitative map of the whole body fat distribution of the measured object;

第一处理单元，用于将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The first processing unit is configured to input the target image into a preset image segmentation model for processing, and segment the subcutaneous fat image and the visceral fat image of each preset part from the target image.
如权利要求8所述的终端设备，其特征在于，所述第一获取单元包括：The terminal device according to claim 8, wherein the first obtaining unit comprises:

第二获取单元，用于获取多个不同回波时间的磁共振图像；a second acquisition unit, configured to acquire a plurality of magnetic resonance images with different echo times;

第一确定单元，用于根据所述多个不同回波时间的磁共振图像确定目标图像。The first determining unit is configured to determine a target image according to the plurality of magnetic resonance images with different echo times.
如权利要求8所述的终端设备，其特征在于，所述第一处理单元包括：The terminal device according to claim 8, wherein the first processing unit comprises:

识别单元，用于从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；The identification unit is used to identify the accurate proton density fat fraction quantitative map corresponding to each preset part from the target image;

第二确定单元，用于基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；a second determining unit, configured to determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

分割单元，用于根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The segmentation unit is configured to segment the subcutaneous fat map and the visceral fat map corresponding to each preset part according to the overall fat map corresponding to each preset part.
一种服务器，包括存储器、处理器以及存储在所述存储器中并可在所述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计算机可读指令时实现如下步骤：A server, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the following steps when executing the computer-readable instructions:

获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；Acquiring a target image; the target image is a proton density fat fraction quantitative map of the whole body fat distribution of the measured object;

将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The target image is input into a preset image segmentation model for processing, and subcutaneous fat images and visceral fat images of each preset part are segmented from the target image.
如权利要求11所述的服务器，其特征在于，所述获取目标图像，包括：The server of claim 11, wherein the acquiring the target image comprises:

获取多个不同回波时间的磁共振图像；Acquiring multiple magnetic resonance images with different echo times;

根据所述多个不同回波时间的磁共振图像确定目标图像。The target image is determined from the plurality of magnetic resonance images with different echo times.
如权利要求11所述的服务器，其特征在于，所述将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图，包括：The server according to claim 11, wherein the target image is input into a preset image segmentation model for processing, and subcutaneous fat images and internal organs of each preset part are segmented from the target image Fat map, including:

从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；Identify the precise proton density fat fraction quantitative map corresponding to each preset part from the target image;

基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；Determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part.
如权利要求13所述的服务器，其特征在于，所述根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The server according to claim 13, wherein the subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part, comprising:

对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；performing a downsampling operation on the overall fat map corresponding to each preset part, and extracting edge features of the total fat map corresponding to each preset part to obtain an edge feature map;

对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图。An up-sampling operation is performed on the edge feature map, the spatial information and edge information in the overall fat map corresponding to each preset part are restored, and a subcutaneous fat map and a visceral fat map corresponding to each preset part are obtained.
如权利要求14所述的服务器，其特征在于，所述对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图，包括：The server according to claim 14, wherein the down-sampling operation is performed on the overall fat map corresponding to each preset part, and edge features of the total fat map corresponding to each preset part are extracted to obtain the edge Feature maps, including:

对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；Perform down-sampling processing on the overall fat map corresponding to each preset part to obtain a first down-sampling feature map;

对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；performing down-sampling processing on the first down-sampling feature map to obtain a second down-sampling feature map;

对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；performing down-sampling processing on the second down-sampling feature map to obtain a third down-sampling feature map;

对所述第三下采样特征图进行下采样处理，得到边缘特征图。Perform down-sampling processing on the third down-sampling feature map to obtain an edge feature map.
如权利要求14所述的服务器，其特征在于，所述对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The server according to claim 14, wherein the upsampling operation is performed on the edge feature map to restore the spatial information and edge information in the overall fat map corresponding to each preset part to obtain each preset part Subcutaneous fat map and visceral fat map corresponding to the site, including:

对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第一通道特征图进行融合，得到第一融合特征图；Perform up-sampling processing on the edge feature map to obtain a first up-sampling feature map, and fuse the first up-sampling feature map with the first channel feature map to obtain a first fused feature map;

对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第二通道特征图进行融合，得到第二融合特征图；Perform up-sampling processing on the first fusion feature map to obtain a second up-sampling feature map, and fuse the second sampling feature map with the second channel feature map to obtain a second fusion feature map;

对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第三通道特征图进行融合，得到第三融合特征图；Perform upsampling processing on the second fusion feature map to obtain a third upsampling feature map, and fuse the third upsampling feature map with the third channel feature map to obtain a third fusion feature map;

对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第四通道特征图进行融合，得到分割结果图。Perform up-sampling processing on the third fusion feature map to obtain a fourth up-sampling feature map, and fuse the fourth up-sampling feature map with the fourth channel feature map to obtain a segmentation result map.
一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，其特征在于，所述计算机可读指令被处理器执行时实现如下步骤：A computer-readable storage medium storing a computer program, wherein the computer-readable instructions are executed by a processor to achieve the following steps:

获取目标图像；所述目标图像为被测对象的全身脂肪分布的质子密度脂肪分数定量图；Acquiring a target image; the target image is a proton density fat fraction quantitative map of the body fat distribution of the subject;

将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图像。The target image is input into a preset image segmentation model for processing, and subcutaneous fat images and visceral fat images of each preset part are segmented from the target image.
如权利要求17所述的计算机可读存储介质，其特征在于，所述获取目标图像，包括：The computer-readable storage medium of claim 17, wherein the acquiring the target image comprises:

获取多个不同回波时间的磁共振图像；Acquiring multiple magnetic resonance images with different echo times;

根据所述多个不同回波时间的磁共振图像确定目标图像。The target image is determined from the plurality of magnetic resonance images with different echo times.
如权利要求17所述的计算机可读存储介质，其特征在于，所述将所述目标图像输入至预设的图像分割模型中进行处理，从所述目标图像中分割出各个预设部位的皮下脂肪图像及内脏脂肪图，包括：The computer-readable storage medium according to claim 17, wherein the target image is input into a preset image segmentation model for processing, and subcutaneous subcutaneous parts of each preset part are segmented from the target image. Fat images and visceral fat maps, including:

从目标图像中识别出各个预设部位对应的精确质子密度脂肪分数定量图；Identify the precise proton density fat fraction quantitative map corresponding to each preset part from the target image;

基于各个预设部位对应的精确质子密度脂肪分数定量图确定出各个预设部位对应的总体脂肪图；Determine the overall fat map corresponding to each preset part based on the accurate proton density fat fraction quantitative map corresponding to each preset part;

根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图。The subcutaneous fat map and the visceral fat map corresponding to each preset part are segmented according to the overall fat map corresponding to each preset part.
如权利要求19所述的计算机可读存储介质，其特征在于，所述根据各个预设部位对应的总体脂肪图分割出各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The computer-readable storage medium according to claim 19, wherein segmenting the subcutaneous fat map and the visceral fat map corresponding to each preset part according to the overall fat map corresponding to each preset part comprises:

对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图；performing a downsampling operation on the overall fat map corresponding to each preset part, and extracting edge features of the total fat map corresponding to each preset part to obtain an edge feature map;

对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图。An upsampling operation is performed on the edge feature map, and the spatial information and edge information in the overall fat map corresponding to each preset part are restored, and the subcutaneous fat map and the visceral fat map corresponding to each preset part are obtained.
如权利要求20所述的计算机可读存储介质，其特征在于，所述对所述各个预设部位对应的总体脂肪图进行下采样操作，提取所述各个预设部位对应的总体脂肪图的边缘特征，得到边缘特征图，包括：The computer-readable storage medium according to claim 20, wherein the downsampling operation is performed on the overall fat map corresponding to each preset part, and the edge of the total fat map corresponding to each preset part is extracted feature to get the edge feature map, including:

对所述各个预设部位对应的总体脂肪图进行下采样处理，得到第一下采样特征图；Perform down-sampling processing on the overall fat map corresponding to each preset part to obtain a first down-sampling feature map;

对所述第一下采样特征图进行下采样处理，得到第二下采样特征图；performing down-sampling processing on the first down-sampling feature map to obtain a second down-sampling feature map;

对所述第二下采样特征图进行下采样处理，得到第三下采样特征图；performing down-sampling processing on the second down-sampling feature map to obtain a third down-sampling feature map;

对所述第三下采样特征图进行下采样处理，得到边缘特征图。Perform down-sampling processing on the third down-sampling feature map to obtain an edge feature map.
如权利要求20所述的计算机可读存储介质，其特征在于，所述对所述边缘特征图进行上采样操作，恢复所述各个预设部位对应的总体脂肪图中的空间信息和边缘信息，得到各个预设部位对应的皮下脂肪图和内脏脂肪图，包括：The computer-readable storage medium according to claim 20, wherein the upsampling operation is performed on the edge feature map to restore the spatial information and edge information in the overall fat map corresponding to each preset part, Obtain the subcutaneous fat map and visceral fat map corresponding to each preset part, including:

对边缘特征图进行上采样处理，得到第一上采样特征图，并将第一上采样特征图与第一通道特征图进行融合，得到第一融合特征图；Perform up-sampling processing on the edge feature map to obtain a first up-sampling feature map, and fuse the first up-sampling feature map with the first channel feature map to obtain a first fused feature map;

对所述第一融合特征图进行上采样处理，得到第二上采样特征图，并将第二采样特征图与第二通道特征图进行融合，得到第二融合特征图；Perform upsampling processing on the first fusion feature map to obtain a second upsampling feature map, and fuse the second sampled feature map with the second channel feature map to obtain a second fusion feature map;

对所述第二融合特征图进行上采样处理，得到第三上采样特征图，并将第三上采样特征图与第三通道特征图进行融合，得到第三融合特征图；Perform upsampling processing on the second fusion feature map to obtain a third upsampling feature map, and fuse the third upsampling feature map with the third channel feature map to obtain a third fusion feature map;

对第三融合特征图进行上采样处理，得到第四上采样特征图，并将第四上采样特征图与第四通道特征图进行融合，得到分割结果图。Perform up-sampling processing on the third fusion feature map to obtain a fourth up-sampling feature map, and fuse the fourth up-sampling feature map with the fourth channel feature map to obtain a segmentation result map.