技术领域technical field
本发明涉及数字图像处理与医学成像的交叉领域，具体地，涉及一种三维精细血管重建方法及其系统。The invention relates to the intersection field of digital image processing and medical imaging, in particular to a three-dimensional fine blood vessel reconstruction method and system thereof.
背景技术Background technique
在生物医学领域，不同的成像技术可以观察到不同的现象和数据，可以综合考虑不同扫描设备的优势，获取真实生物软组织多种尺度下的细微观图像数据。CT图像是对人体某一部分的扫描图像，可以对血管、肿瘤等组织成像，虽然成像完整，但其缺点在于扫描数据的细节质量明显不如组织切片，而且能分辨的组织类型有限。利用光学和电子显微镜对组织切片成像，能够获取分辨清晰度更高、组织结构类型更丰富的细微观多尺度图像数据，但获得的图像数据易存在局部区域缺失的问题，因此，需要对高精度扫描图像中的局部缺失区域进行修复重建。In the field of biomedicine, different imaging technologies can observe different phenomena and data, and the advantages of different scanning equipment can be considered comprehensively to obtain microscopic image data of real biological soft tissues at multiple scales. A CT image is a scanned image of a certain part of the human body. It can image blood vessels, tumors and other tissues. Although the imaging is complete, its disadvantage is that the quality of details of the scanned data is obviously not as good as that of tissue slices, and the types of tissues that can be distinguished are limited. Using optical and electron microscopes to image tissue slices can obtain fine-scale multi-scale image data with higher resolution and richer tissue structure types, but the obtained image data is prone to missing local areas. Therefore, high-precision Local missing regions in scanned images are repaired and reconstructed.
医学中的三维精细血管图像有着类似树形的复杂拓扑，尤其血管分支上存在分叉区域，而且分叉区域的分支数量多少不一，同时血管的粗细尺度变化大，在可视化研究中，对血管的体绘制一直是个棘手的问题，而外科手术中对主要的血管形态需要有准确的描述，以辅助医生及时选择合理治疗方法，比如从血管的三维结构上观察和确定病变位置，做出快速诊断；为手术计划提供直观的参考依据。因此一旦三维精细血管图像出现局部缺失，就会影响医生的诊断。The three-dimensional fine blood vessel image in medicine has a complex tree-like topology, especially the bifurcation area on the branch of the blood vessel, and the number of branches in the bifurcation area is different, and the thickness of the blood vessel varies greatly. Volume rendering has always been a thorny problem, and the main vascular morphology needs to be accurately described in surgery to assist doctors in choosing reasonable treatment methods in a timely manner, such as observing and determining the location of lesions from the three-dimensional structure of blood vessels, and making a rapid diagnosis ; Provide intuitive reference basis for operation planning. Therefore, once the three-dimensional fine blood vessel image is partially missing, it will affect the doctor's diagnosis.
所以，有必要提供一种针对有缺失部位的高精度的三维精细血管图像进行重建的技术方案。Therefore, it is necessary to provide a technical solution for reconstructing high-precision three-dimensional fine blood vessel images with missing parts.
发明内容Contents of the invention
本发明提供一种三维精细血管重建方法及其系统，通过映射位置关系和计算整体匹配度获取重建缺失部位的填充区域，通过填充区域修复缺失部位获得完整的高精度图像，三维精细血管重建的精确性高。The present invention provides a three-dimensional fine blood vessel reconstruction method and its system. The filling area of the missing part is obtained by mapping the positional relationship and calculating the overall matching degree, and a complete high-precision image is obtained by repairing the missing part through the filling area. The accuracy of the three-dimensional fine blood vessel reconstruction is accurate. high sex.
一种三维精细血管重建方法，包括以下步骤：A three-dimensional fine blood vessel reconstruction method, comprising the following steps:
步骤一、载入三维精细血管原始图像：Step 1. Load the original image of 3D fine blood vessels:
所述三维精细血管原始图像包括扫描同一对象的待修复图像和参考图像，所述待修复图像为具缺失部位的血管图像，所述参考图像为完整的血管图像，所述待修复图像的精度高于所述参考图像的精度；The three-dimensional fine blood vessel original image includes an image to be repaired and a reference image of the same object scanned, the image to be repaired is a blood vessel image with a missing part, the reference image is a complete blood vessel image, and the image to be repaired has high precision based on the accuracy of the reference image;
根据所述待修复图像指定待修复区域，并定义包含所述待修复区域的待修复块，根据所述待修复块确定在所述参考图像中位置相应、大小相同的参考映射块，所述待修复块中的待修复区域对应所述参考映射块中的参考映射区域；Designate an area to be repaired according to the image to be repaired, and define a block to be repaired containing the area to be repaired, and determine a reference mapping block with a corresponding position and the same size in the reference image according to the block to be repaired, and the block to be repaired is determined according to the block to be repaired. The area to be repaired in the repair block corresponds to the reference mapping area in the reference mapping block;
步骤二、基于血管特性的图像预处理：Step 2. Image preprocessing based on blood vessel characteristics:
对载入的所述三维精细血管原始图像进行分析，确定需要增强的血管区域，对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像，所述三维精细血管增强图像包括增强后的待修复图像和增强后的参考图像；Analyzing the loaded original image of the three-dimensional fine blood vessel, determining the area of the blood vessel that needs to be enhanced, and enhancing the original image of the three-dimensional fine blood vessel to obtain an enhanced image of the three-dimensional fine blood vessel, the enhanced image of the three-dimensional fine blood vessel includes an enhanced The image to be repaired and the enhanced reference image;
步骤三、基于血管特性的特征匹配：Step 3. Feature matching based on blood vessel characteristics:
根据所述三维精细血管增强图像，计算整体匹配度，得到所述参考映射块在所述待修复图像中的最优匹配块，所述参考映射块中的参考映射区域对应所述最优匹配块中的填充区域；Calculate the overall matching degree according to the three-dimensional fine blood vessel enhanced image to obtain the optimal matching block of the reference mapping block in the image to be repaired, and the reference mapping area in the reference mapping block corresponds to the optimal matching block Filled area in ;
步骤四、图像填充及边界处理：Step 4, image filling and boundary processing:
将所述最优匹配块的填充区域对应填充至所述待修复块的待修复区域，并对边界进行平滑处理，完成修复重建。Correspondingly filling the filled area of the optimal matching block to the area to be repaired of the block to be repaired, and smoothing the boundary to complete the restoration and reconstruction.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述步骤二包括如下步骤：In a preferred embodiment of the three-dimensional fine vessel reconstruction method provided by the present invention, the second step includes the following steps:
基于Hessian矩阵对所述三维精细血管原始图像进行分析：选取Hessian矩阵对所述三维精细血管原始图像做卷积运算，求得特征值和特征向量；Analyzing the original image of the three-dimensional fine blood vessel based on the Hessian matrix: selecting the Hessian matrix to perform a convolution operation on the original image of the three-dimensional fine blood vessel to obtain eigenvalues and eigenvectors;
建立血管区域特征函数：基于血管特性并根据所述特征值和特征向量建立血管区域特征函数；Establishing the characteristic function of the blood vessel area: establishing the characteristic function of the blood vessel area based on the characteristics of the blood vessel and according to the eigenvalue and the eigenvector;
确定需要增强的血管区域：根据所述血管区域特征函数确定需要增强的血管区域，进而对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像。Determining the blood vessel region that needs to be enhanced: determining the blood vessel region that needs to be enhanced according to the characteristic function of the blood vessel region, and then enhancing the original three-dimensional fine blood vessel image to obtain a three-dimensional fine blood vessel enhanced image.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述基于Hessian矩阵对所述三维精细血管原始图像进行分析的步骤包括如下：In a preferred embodiment of the three-dimensional fine blood vessel reconstruction method provided by the present invention, the step of analyzing the original image of the three-dimensional fine blood vessel based on the Hessian matrix includes the following steps:
计算所述三维精细血管原始图像中各像素点的二阶偏导数和混合偏导数；calculating the second-order partial derivative and mixed partial derivative of each pixel in the original three-dimensional fine blood vessel image;
以计算的二阶偏导数和混合偏导数组成Hessian矩阵与所述三维精细血管原始图像做卷积运算；Convolving the Hessian matrix with the calculated second-order partial derivatives and mixed partial derivatives with the original image of the three-dimensional fine blood vessels;
求得特征值λ1、λ2及λ3和与所述特征值对应的特征向量γ1、γ2及γ3，且|λ1|＜|λ2|＜|λ3|。Eigenvalues λ1 , λ2 , and λ3 and eigenvectors γ1 , γ2 , and γ3 corresponding to the eigenvalues are obtained, and |λ1 |<|λ2 |<|λ3 |.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述血管区域特征函数为：In a preferred embodiment of the three-dimensional fine blood vessel reconstruction method provided by the present invention, the characteristic function of the blood vessel region is:
定义RA＝|λ2|/|λ3|，其中，a、c以及d分别为平面结构参数、背景区分参数以及三维血管结构参数，k1、k2及k3分别为沿所述特征向量γ1、γ2、γ3反方向的梯度方向的灰度变化率的大小。Define RA =|λ2 |/|λ3 |, Wherein, a, c, and d are plane structure parameters, background discrimination parameters, and three-dimensional blood vessel structure parameters, respectively, and k1 , k2 , and k3 are gradient directions along the opposite directions of the feature vectors γ1 , γ2 , and γ3 respectively. The size of the grayscale change rate.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述步骤三包括如下步骤：In a preferred embodiment of the three-dimensional fine blood vessel reconstruction method provided by the present invention, the third step includes the following steps:
将所述待修复图像均分为多个与所述参考映射块大小相同的候选块，计算所述候选块和所述参考映射块的全局匹配度SM；Dividing the image to be repaired into multiple candidate blocks with the same size as the reference mapping block, and calculating the global matching SM of the candidate block and the reference mapping block;
计算血管段相似度ST；Calculate the similarity ST of the blood vessel segment;
根据所述全局匹配度SM和所述血管段相似度ST计算整体匹配度Sves，所述整体匹配度Sves＝μSM+ηST，其中，μ及η分别为SM及ST的权重系数，且μ+η＝1；Calculate the overall matching degree Sves according to the global matching degree SM and the blood vessel segment similarity ST , the overall matching degree Sves = μSM +ηST , wherein μ and η are SM andST respectively The weight coefficient of , and μ+η=1;
根据所述整体匹配度Sves，得到所述参考映射块在所述待修复图像中的最优匹配块，所述参考映射块中的参考映射区域对应所述最优匹配块中的填充区域。According to the overall matching degree Sves , an optimal matching block of the reference mapping block in the image to be repaired is obtained, and a reference mapping area in the reference mapping block corresponds to a filling area in the optimal matching block.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述计算血管段相似度ST的步骤包括如下步骤：In a preferred embodiment of the three-dimensional fine blood vessel reconstruction method provided by the present invention, the step of calculating the similarityST of the blood vessel segment includes the following steps:
基于逐层剥取法细化步骤二中所述需要增强的血管区域，提取单像素中心线，由所述三维精细血管增强图像得到三维精细血管二值图，所述三维精细血管二值图包括所述待修复图像的二值图和所述参考图像的二值图；Based on the layer-by-layer stripping method, the vascular area to be enhanced is refined in step 2, and the single-pixel centerline is extracted, and the 3D fine vascular binary image is obtained from the 3D fine vascular enhanced image, and the 3D fine vascular binary image includes all Describe the binary image of the image to be repaired and the binary image of the reference image;
根据所述待修复图像的二值图提取所述候选块的端点集，根据所述参考图像的二值图提取所述参考映射块的端点集，分别求取端点特征向量；Extracting the endpoint set of the candidate block according to the binary image of the image to be repaired, extracting the endpoint set of the reference mapping block according to the binary image of the reference image, and obtaining the endpoint feature vectors respectively;
根据所述端点特征向量计算血管段相似度ST。The blood vessel segment similarity ST is calculated according to the endpoint feature vector.
在本发明提供的三维精细血管重建方法的一种较佳实施例中，所述步骤四包括如下步骤：In a preferred embodiment of the three-dimensional fine vessel reconstruction method provided by the present invention, the fourth step includes the following steps:
将所述最优匹配块的填充区域对应填充至所述待修复块的待修复区域；Correspondingly filling the filling area of the optimal matching block to the area to be repaired of the block to be repaired;
将所述填充区域的边界进行插值迭代，不断逼近直至收敛，将所述填充区域、以及所述待修复区域的边界平滑过渡，完成修复重建。Interpolation iteration is performed on the boundary of the filling area until convergence, and the boundary of the filling area and the area to be repaired is smoothly transitioned to complete the restoration and reconstruction.
本发明还提供一种三维精细血管重建系统，包括图像载入模块，用于载入三维精细血管原始图像，所述三维精细血管原始图像包括扫描同一对象的待修复图像和参考图像，所述待修复图像为具缺失的血管图像，所述参考图像为完整的血管图像，所述待修复图像的精度高于所述参考图像；并根据所述待修复图像指定待修复区域，并定义包含所述待修复区域的待修复块，根据所述待修复块确定在所述参考图像中位置相应、大小相同的参考映射块，所述待修复块中的待修复区域对应所述参考映射块中的参考映射区域；The present invention also provides a three-dimensional fine blood vessel reconstruction system, which includes an image loading module for loading the original image of the three-dimensional fine blood vessel, the original image of the three-dimensional fine blood vessel includes an image to be repaired and a reference image that scan the same object, the to-be-repaired image The repaired image is a missing blood vessel image, the reference image is a complete blood vessel image, and the precision of the image to be repaired is higher than that of the reference image; and the region to be repaired is specified according to the image to be repaired, and the definition includes the The block to be repaired in the area to be repaired, according to the block to be repaired, a reference mapping block with a corresponding position and the same size in the reference image is determined, and the area to be repaired in the block to be repaired corresponds to the reference in the reference mapping block map area;
图像预处理模块，用于对所述图像载入模块载入的所述三维精细血管原始图像进行分析，确定需要增强的血管区域，对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像，所述三维精细血管增强图像包括增强后的待修复图像和增强后的参考图像；An image preprocessing module, configured to analyze the original image of the three-dimensional fine blood vessel loaded by the image loading module, determine the area of the blood vessel that needs to be enhanced, and enhance the original image of the three-dimensional fine blood vessel to obtain the enhanced three-dimensional fine blood vessel image, the three-dimensional fine blood vessel enhanced image includes an enhanced image to be repaired and an enhanced reference image;
特征匹配模块，用于根据所述图像预处理模块得到的三维精细血管增强图像，计算整体匹配度，得到所述参考映射块在所述待修复图像中的最优匹配块，所述参考映射块中的参考映射区域对应所述最优匹配块中的填充区域；The feature matching module is used to calculate the overall matching degree according to the three-dimensional fine blood vessel enhanced image obtained by the image preprocessing module, and obtain the optimal matching block of the reference mapping block in the image to be repaired, and the reference mapping block The reference mapping area in corresponds to the padding area in the optimal matching block;
填充处理模块，用于将所述特征匹配模块得到的所述最优匹配块的填充区域对应填充至所述待修复块的待修复区域，并对边界进行平滑处理，完成修复重建。The filling processing module is used to fill the filling area of the optimal matching block obtained by the feature matching module into the area to be repaired of the block to be repaired, and smooth the boundary to complete the restoration and reconstruction.
在本发明提供的三维精细血管重建系统的一种较佳实施例中，所述图像预处理模块包括：In a preferred embodiment of the three-dimensional fine vessel reconstruction system provided by the present invention, the image preprocessing module includes:
第一运算单元，选取Hessian矩阵对所述三维精细血管原始图像做卷积运算，求得特征值和特征向量；The first operation unit selects the Hessian matrix to perform convolution operation on the three-dimensional fine blood vessel original image to obtain eigenvalues and eigenvectors;
函数创建单元，根据所述第一运算单元求得的特征值和特征向量并基于血管特性建立血管区域特征函数；A function creation unit, based on the eigenvalues and eigenvectors obtained by the first operation unit and based on the characteristics of blood vessels to establish a vascular region feature function;
血管增强单元，根据所述函数创建单元建立的血管区域特征函数确定需要增强的血管区域，进而对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像。The blood vessel enhancement unit determines the blood vessel region to be enhanced according to the blood vessel region feature function established by the function creation unit, and then enhances the original three-dimensional fine blood vessel image to obtain a three-dimensional fine blood vessel enhanced image.
在本发明提供的三维精细血管重建系统的一种较佳实施例中，所述特征匹配模块包括第一计算单元、第二计算单元及第三计算单元，分别用于计算全局匹配度、血管段相似度及整体匹配度。In a preferred embodiment of the three-dimensional fine blood vessel reconstruction system provided by the present invention, the feature matching module includes a first calculation unit, a second calculation unit and a third calculation unit, which are used to calculate the global matching degree, blood vessel segment similarity and overall matching.
相较于现有技术，本发明提供的三维精细血管重建方法及其系统，具有以下有益效果：Compared with the prior art, the three-dimensional fine vessel reconstruction method and system provided by the present invention have the following beneficial effects:
一、本发明提供的三维精细血管重建方法对需要增强的血管区域进行了特征值提取，将三维精细血管原始图像进行了图像增强，便于对源于同一血管样本的待修复图像和参考图像进行结构匹配，提高了三维精细血管重建的效率，并能够保证精度；1. The three-dimensional fine blood vessel reconstruction method provided by the present invention extracts the feature value of the blood vessel area that needs to be enhanced, and performs image enhancement on the original image of the three-dimensional fine blood vessel, so as to facilitate the structure of the image to be repaired and the reference image originating from the same blood vessel sample. Matching improves the efficiency of three-dimensional fine vessel reconstruction and ensures accuracy;
二、本发明重新定义了基于Hessian矩阵的血管区域特征函数，考虑了三维精细血管原始图像的灰度信息，可以有效提高图像增强的效果；同时考虑了血管的分叉区域以及弧度较大的区域等个体差异性的影响，能够很好的解决个体差异性的问题，即使在特殊情况也不会判断错误，具有更广泛的适用性；2. The present invention redefines the characteristic function of the blood vessel area based on the Hessian matrix, and considers the gray information of the original image of the three-dimensional fine blood vessel, which can effectively improve the effect of image enhancement; at the same time, the bifurcation area of the blood vessel and the area with a large arc are considered Such as the influence of individual differences, can solve the problem of individual differences well, even in special cases, it will not make mistakes in judgment, and has wider applicability;
三、本发明提供的三维精细血管重建方法基于血管特性而成，考虑到血管结构的管状特性和较强的自相似性，从而指导源于同一血管样本的待修复图像和参考图像的匹配，提高了血管结构特征匹配的可靠性和准确性，进一步增大三维精细血管重建的精度；3. The three-dimensional fine blood vessel reconstruction method provided by the present invention is based on the characteristics of blood vessels, taking into account the tubular characteristics and strong self-similarity of the blood vessel structure, so as to guide the matching of the image to be repaired and the reference image from the same blood vessel sample, and improve Improve the reliability and accuracy of vascular structure feature matching, and further increase the accuracy of 3D fine vessel reconstruction;
四、本发明通过待修复图像和参考图像的匹配，得到参考映射块在待修复图像中的最优匹配块，对应填充之后再对边界进行平滑处理，修复精度高。4. The present invention obtains the optimal matching block of the reference mapping block in the image to be repaired by matching the image to be repaired and the reference image, and smoothes the boundary after the corresponding filling, so that the repair accuracy is high.
附图说明Description of drawings
图1为本发明提供的三维精细血管重建方法的修复原理图；Fig. 1 is a schematic diagram of the repair principle of the three-dimensional fine blood vessel reconstruction method provided by the present invention;
图2为本发明提供的三维精细血管重建方法的流程图；Fig. 2 is a flow chart of the three-dimensional fine blood vessel reconstruction method provided by the present invention;
图3(a)为本发明提供的三维精细血管原始图像中的参考图像；Fig. 3 (a) is the reference image in the three-dimensional fine blood vessel original image provided by the present invention;
图3(b)为本发明提供的三维精细血管原始图像中的待修复图像；Fig. 3 (b) is the image to be repaired in the three-dimensional fine blood vessel original image provided by the present invention;
图4为图2所示三维精细血管重建方法中步骤S2的流程图；Fig. 4 is a flowchart of step S2 in the three-dimensional fine vessel reconstruction method shown in Fig. 2;
图5为图4所示步骤S2中步骤S21的流程图；Fig. 5 is the flowchart of step S21 in step S2 shown in Fig. 4;
图6(a)为图3(a)所示增强后的参考图像；Figure 6(a) is the enhanced reference image shown in Figure 3(a);
图6(b)为图3(b)所示增强后的待修复图像；Figure 6(b) is the enhanced image to be repaired shown in Figure 3(b);
图7为图2所示三维精细血管重建方法中步骤S3的流程图；Fig. 7 is a flowchart of step S3 in the three-dimensional fine vessel reconstruction method shown in Fig. 2;
图8为图7所示步骤S3中步骤S32的流程图；Fig. 8 is the flowchart of step S32 in step S3 shown in Fig. 7;
图9(a)为图6(a)所示细化后的参考图像；Figure 9(a) is the reference image after refinement shown in Figure 6(a);
图9(b)为图6(b)所示细化后的待修复图像；Figure 9(b) is the image to be repaired after refinement shown in Figure 6(b);
图10为图2所示三维精细血管重建方法中步骤S4的流程图；Fig. 10 is a flowchart of step S4 in the three-dimensional fine vessel reconstruction method shown in Fig. 2;
图11(a)为待修复的一种三维精细血管结构图；Figure 11(a) is a three-dimensional fine blood vessel structure diagram to be repaired;
图11(b)为图11(a)所示修复后的三维精细血管结构图；Figure 11(b) is a three-dimensional fine blood vessel structure diagram after repair shown in Figure 11(a);
图12为本发明提供的三维精细血管重建系统的结构框图；Fig. 12 is a structural block diagram of the three-dimensional fine vessel reconstruction system provided by the present invention;
图13为图12所示图像预处理模块的结构框图；Fig. 13 is a structural block diagram of the image preprocessing module shown in Fig. 12;
图14为图12所示特征匹配模块的结构框图；Fig. 14 is a structural block diagram of the feature matching module shown in Fig. 12;
图15为图12所示填充处理模块的结构框图。Fig. 15 is a structural block diagram of the filling processing module shown in Fig. 12 .
具体实施方式detailed description
下面将结合附图和实施方式对本实用新型(发明)作进一步说明。请参阅图1，为本发明提供的三维精细血管重建方法的修复原理图。本发明需解决的技术问题是修复具缺失部位的三维精细血管图像，记为待修复图像1。本发明还提供了用于修复所述待修复图像1的参考图像3。具体地，在不破坏血管样本的前提下，为了获得血管样本的微观多尺度图像数据，对一份血管样本进行序列连续切片，采用分辨率高的光学和电子显微镜进行扫描获得所述待修复图像1，由于分辨率越高，能扫描样本区域的范围越小，对血管样本进行细微观多尺度扫描获得的图像数据存在缺失部位，即所述待修复图像1为具缺失部位的高精度血管图像；采用显微CT/MRI和同步辐射CT直接扫描同一血管样本，获得所述参考图像3，所述参考图像3为完整的低精度血管图像；此低精度和高精度是相对概念，即所述待修复图像1的扫描精度高于所述参考图像3的扫描精度。The utility model (invention) will be further described below in conjunction with the accompanying drawings and embodiments. Please refer to FIG. 1 , which is a repair schematic diagram of the three-dimensional fine vessel reconstruction method provided by the present invention. The technical problem to be solved in the present invention is to repair the three-dimensional fine blood vessel image with missing parts, which is denoted as image to be repaired 1 . The present invention also provides a reference image 3 for repairing the image 1 to be repaired. Specifically, on the premise of not destroying the blood vessel sample, in order to obtain the microscopic multi-scale image data of the blood vessel sample, a blood vessel sample is serially sliced, and the image to be repaired is obtained by scanning with a high-resolution optical and electron microscope 1. Since the higher the resolution, the smaller the range of the sample area that can be scanned, the image data obtained by microscopic multi-scale scanning of the blood vessel sample has missing parts, that is, the image to be repaired 1 is a high-precision blood vessel image with missing parts ; The same blood vessel sample is directly scanned by micro-CT/MRI and synchrotron radiation CT to obtain the reference image 3, which is a complete low-precision blood vessel image; this low precision and high precision are relative concepts, that is, the The scanning accuracy of the image to be repaired 1 is higher than that of the reference image 3 .
T1、指定所述待修复图像1中所述缺失部位为待修复区域10，定义从所述待修复区域10向外延伸获得的立方体扩张区域为待修复块12；T1. Designate the missing part in the image to be repaired 1 as the region to be repaired 10, and define the cube expansion area obtained by extending outward from the region to be repaired 10 as the block to be repaired 12;
T2、通过映射位置关系在所述参考图像3中获得与所述待修复块12位置相应、大小相同的参考映射块32，所述待修复块12中的待修复区域10对应所述参考映射块32中的参考映射区域30；T2. Obtain a reference mapping block 32 corresponding to the position of the block to be repaired 12 and having the same size in the reference image 3 through the mapping position relationship, and the region to be repaired 10 in the block to be repaired 12 corresponds to the reference mapping block reference map area 30 in 32;
T3、在所述待修复图像1中搜索匹配与所述参考映射块32相似的最优匹配块52，再通过映射位置关系获得与所述待修复区域10相对应的填充区域50；T3. Search and match the optimal matching block 52 similar to the reference mapping block 32 in the image to be repaired 1, and then obtain the filling area 50 corresponding to the area to be repaired 10 through the mapping position relationship;
T4、将所述填充区域50对应填充至所述待修复区域10，所述填充区域50修复所述缺失部位，对所述填充区域50和所述待修复区域10的边界进行平滑处理，得到自然过渡的纹理图像，完成修复重建，得到完整的高精度三维精细血管图像。T4. Correspondingly fill the filled area 50 to the area to be repaired 10, the filled area 50 repairs the missing part, and smooth the boundary between the filled area 50 and the area to be repaired 10 to obtain a natural The transitional texture image is repaired and reconstructed to obtain a complete high-precision three-dimensional fine blood vessel image.
请参阅图2，为本发明提供的三维精细血管重建方法的流程图。本发明提供一种三维精细血管重建方法，包括如下步骤：Please refer to FIG. 2 , which is a flowchart of the three-dimensional fine vessel reconstruction method provided by the present invention. The present invention provides a three-dimensional fine blood vessel reconstruction method, comprising the following steps:
步骤S1、载入三维精细血管原始图像：Step S1. Load the original image of the three-dimensional fine blood vessels:
所述三维精细血管原始图像包括扫描同一对象的待修复图像1和参考图像3，所述待修复图像1为具缺失部位的血管图像，所述参考图像3为完整的血管图像，所述待修复图像1的精度高于所述参考图像3的精度。The three-dimensional fine blood vessel original image includes an image 1 to be repaired and a reference image 3 that scan the same object, the image 1 to be repaired is a blood vessel image with missing parts, the reference image 3 is a complete blood vessel image, and the image to be repaired The accuracy of image 1 is higher than that of the reference image 3 .
在本实施例中，具体地，请参见图3(a)及图3(b)，其中，图3(a)为本发明提供的三维精细血管原始图像中的参考图像；图3(b)为本发明提供的三维精细血管原始图像中的待修复图像。需要说明的是，为了更清楚的描述本发明提供的方法中的关键步骤，本发明对上述源于同一血管样本的参考图像和待修复图像进行了适当截取，并以此为实例进行说明，但不局限于上述公开的图像。In this embodiment, specifically, please refer to Fig. 3(a) and Fig. 3(b), wherein, Fig. 3(a) is a reference image in the original image of three-dimensional fine blood vessels provided by the present invention; Fig. 3(b) It is the image to be repaired in the 3D fine blood vessel original image provided by the present invention. It should be noted that, in order to more clearly describe the key steps in the method provided by the present invention, the present invention appropriately intercepts the above-mentioned reference image and the image to be repaired from the same blood vessel sample, and uses this as an example for illustration, but Not limited to the images disclosed above.
根据所述待修复图像1指定待修复区域10，并定义包含所述待修复区域10的待修复块12，根据所述待修复块12确定在所述参考图像3中位置相应、大小相同的参考映射块32，所述待修复块12中的待修复区域10对应所述参考映射块32中的参考映射区域30。Designate the area to be repaired 10 according to the image to be repaired 1, and define the block to be repaired 12 containing the area to be repaired 10, and determine the reference image 3 with the corresponding position and the same size according to the block to be repaired 12 In the mapping block 32 , the area to be repaired 10 in the block to be repaired 12 corresponds to the reference mapping area 30 in the reference mapping block 32 .
在本实施例中，指定所述待修复图像1中所述缺失部位为待修复区域10，定义从所述待修复区域10向外延伸获得的立方体扩张区域为待修复块12。通过映射位置关系在所述参考图像3中获得与所述待修复块12位置相应、大小相同的参考映射块32，所述待修复块12中的待修复区域10对应所述参考映射块32中的参考映射区域30。In this embodiment, the missing part in the image to be repaired 1 is designated as the region to be repaired 10 , and the cubic expansion area obtained by extending outward from the region to be repaired 10 is defined as the block to be repaired 12 . Obtain a reference mapping block 32 corresponding to the position of the block to be repaired 12 and having the same size in the reference image 3 through the mapping position relationship, and the area to be repaired 10 in the block to be repaired 12 corresponds to the reference mapping block 32 The reference map area 30.
步骤S2、基于血管特性的图像预处理：Step S2, image preprocessing based on blood vessel characteristics:
对载入的所述三维精细血管原始图像进行分析，确定需要增强的血管区域，对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像，所述三维精细血管增强图像包括增强后的待修复图像1和增强后的参考图像3。Analyzing the loaded original image of the three-dimensional fine blood vessel, determining the area of the blood vessel that needs to be enhanced, and enhancing the original image of the three-dimensional fine blood vessel to obtain an enhanced image of the three-dimensional fine blood vessel, the enhanced image of the three-dimensional fine blood vessel includes an enhanced Image 1 to be repaired and reference image 3 after enhancement.
步骤S3、基于血管特性的特征匹配：Step S3, feature matching based on blood vessel characteristics:
根据所述三维精细血管增强图像，计算整体匹配度，得到所述参考映射块32在所述待修复图像1中的最优匹配块52，所述参考映射块32中的参考映射区域30对应所述最优匹配块52中的填充区域50。According to the three-dimensional fine blood vessel enhanced image, the overall matching degree is calculated to obtain the optimal matching block 52 of the reference mapping block 32 in the image to be repaired 1, and the reference mapping area 30 in the reference mapping block 32 corresponds to the Filled area 50 in optimal matching block 52 described above.
步骤S4、图像填充及边界处理：Step S4, image filling and boundary processing:
将所述最优匹配块52的填充区域50对应填充至所述待修复块12的待修复区域10，并对边界进行平滑处理，完成修复重建。The filling area 50 of the optimal matching block 52 is correspondingly filled to the area to be repaired 10 of the block to be repaired 12, and the boundary is smoothed to complete the restoration and reconstruction.
请参阅图4，为图2所示三维精细血管重建方法中步骤S2的流程图。由于采集设备光强的误差，三维精细血管等医学图像中感兴趣部位(即血管区域)的灰度值可能并不明显，为了便于对图像进行修复，需要先对图像进行增强处理，增强关键结构的匹配。Please refer to FIG. 4 , which is a flowchart of step S2 in the three-dimensional fine vessel reconstruction method shown in FIG. 2 . Due to the error of the light intensity of the acquisition equipment, the gray value of the interesting part (that is, the blood vessel area) in the three-dimensional fine blood vessel and other medical images may not be obvious. In order to facilitate image restoration, the image needs to be enhanced first to enhance the key structure. match.
根据结构张量能够识别图像中的边缘、角点以及平坦区域，结构张量是对图像中的像素利用矩阵组织的数据结构，其形式便是Hessian矩阵。对于形状特征特殊的血管图像，血管截面的灰度成像高斯分布的特性，可以利用结构张量有效得到血管所需特征，进而对其进行特征增强。According to the structure tensor, the edges, corners and flat areas in the image can be identified. The structure tensor is a data structure organized by a matrix for the pixels in the image, and its form is the Hessian matrix. For blood vessel images with special shape features, the characteristics of Gaussian distribution of gray-scale imaging of blood vessel sections can use the structure tensor to effectively obtain the required characteristics of blood vessels, and then enhance their features.
所述基于血管特性的图像预处理步骤S2包括：The image preprocessing step S2 based on blood vessel characteristics includes:
步骤S21、基于Hessian矩阵对所述三维精细血管原始图像进行分析：选取Hessian矩阵对所述三维精细血管原始图像做卷积运算，求得特征值和特征向量。Step S21 , analyzing the original image of the three-dimensional fine blood vessel based on the Hessian matrix: performing convolution operation on the original image of the three-dimensional fine blood vessel by selecting the Hessian matrix to obtain eigenvalues and eigenvectors.
请同时参阅图5，为图4所示步骤S2中步骤S21的流程图。Please also refer to FIG. 5 , which is a flowchart of step S21 in step S2 shown in FIG. 4 .
所述基于Hessian矩阵对所述三维精细血管原始图像进行分析的步骤S21包括：The step S21 of analyzing the original three-dimensional fine blood vessel image based on the Hessian matrix includes:
步骤S21-1、计算所述三维精细血管原始图像中各像素点的二阶偏导数和混合偏导数；Step S21-1, calculating the second-order partial derivative and mixed partial derivative of each pixel in the original image of the three-dimensional fine blood vessel;
步骤S21-2、以计算的二阶偏导数和混合偏导数组成Hessian矩阵与所述三维精细血管原始图像做卷积运算；Step S21-2, using the calculated second-order partial derivatives and mixed partial derivatives to form a Hessian matrix and perform a convolution operation with the original image of the three-dimensional fine blood vessels;
步骤S21-3、求得特征值λ1、λ2及λ3和与所述特征值对应的特征向量γ1、γ2及γ3，且|λ1|＜|λ2|＜|λ3|。Step S21-3, obtain the eigenvalues λ1 , λ2 and λ3 and the eigenvectors γ1 , γ2 and γ3 corresponding to the eigenvalues, and |λ1 |<|λ2 |<|λ3 |.
具体过程为：The specific process is:
设Vx,Vy,Vz分别为x,y,z方向的偏导数，计算所述三维精细血管原始图像中各像素点的二阶偏导数和混合偏导数，其中Vσ＝Gσ*V，*代表卷积算子，通过高斯函数(1)获得不同尺度σ中的线性增强滤波。进行局部特性分析时，以当前处理像素点为中心，在当前所处理的图像数据上，取半宽为3σ的矩形窗口，足以包含血管的直径。Let Vx , Vy , and Vz be the partial derivatives in the x, y, and z directions respectively, Calculate the second-order partial derivatives and mixed partial derivatives of each pixel in the three-dimensional fine blood vessel original image, where Vσ =Gσ *V, * represents the convolution operator, and the Gaussian function (1) is used to obtain the Linear enhancement filtering. When performing local characteristic analysis, take the currently processed pixel as the center and take a rectangular window with a half-width of 3σ on the currently processed image data, which is sufficient to include the diameter of the blood vessel.
所述高斯函数(1)为：Described Gaussian function (1) is:
结构张量(是一个三维的Hessian矩阵)定义如下式(2)：structure tensor (is a three-dimensional Hessian matrix) defined as follows (2):
其中是笛卡尔内积；存在一个三维正交矩阵S，使得其中Λ＝diag(λm)的对角元素是的特征值，组成矩阵S的各行向量即为结构张量Υ的特征向量，分别记为λ1、λ2及λ3，且满足|λ1|＜|λ2|＜|λ3|，与三个特征值分别一一对应的三个特征向量记为γ1、γ2及γ3，其中γ1⊥λ2,γ1⊥λ3,γ2⊥λ3。结构张量Υ的特征值λ1、λ2及λ3能够体现出图像灰度在相应特征向量方向上的变化速度。图像的边缘强度大，可以量化为结构张量Υ较大的特征值。in is the Cartesian inner product; there exists a three-dimensional orthogonal matrix S such that where the diagonal elements of Λ=diag(λm ) are The eigenvalues of the matrix S, the row vectors of the matrix S are the eigenvectors of the structure tensor Y, denoted as λ1 , λ2 and λ3 respectively, and satisfy |λ1 |<|λ2 |<|λ3 |, and The three eigenvectors corresponding to the three eigenvalues are denoted as γ1 , γ2 and γ3 , where γ1 ⊥λ2 , γ1 ⊥λ3 , and γ2 ⊥λ3 . The eigenvalues λ1 , λ2 and λ3 of the structure tensor Y can reflect the change speed of the image gray level in the direction of the corresponding eigenvector. The edge strength of the image is large, which can be quantized as the eigenvalue with large structure tensor Υ.
根据三个特征值|λ1|＜|λ2|＜|λ3|，对矩阵的特征值及对应的形状结构进行如下分析：According to the three eigenvalues |λ1 |<|λ2 |<|λ3 |, the eigenvalues and corresponding shape structures of the matrix are analyzed as follows:
a.|λ1|≈0,|λ2|≈0,|λ3|≈0,可判定为平坦区域；可以判定为面状结构；可以判定为管状结构；可判定为球状结构。a.|λ1 |≈0,|λ2 |≈0,|λ3 |≈0, can be judged as a flat area; Can be judged as planar structure; It can be judged as a tubular structure; It can be judged as spherical structure.
通过上述分析，对于理想的三维精细血管图像，其特征值及特征向量的关系应该如下式(3)：Through the above analysis, for an ideal three-dimensional fine blood vessel image, the relationship between its eigenvalues and eigenvectors should be as follows (3):
其中λ2,λ3均为负数，表示了在三维精细血管图像中感兴趣部位为高亮区而背景为阴暗区；矩阵的两个特征向量γ2,γ3组成的面所代表的是三维精细血管的切面，而最小特征值所对应的特征向量γ1代表的是三维精细血管的走向，即三维精细血管的延伸方向。Among them, λ2 and λ3 are both negative numbers, which means that in the three-dimensional fine blood vessel image, the part of interest is a bright area and the background is a dark area; the surface composed of two eigenvectors γ2 and γ3 of the matrix represents a three-dimensional The cut plane of the fine blood vessels, and the eigenvector γ1 corresponding to the minimum eigenvalue represents the direction of the three-dimensional fine blood vessels, that is, the extension direction of the three-dimensional fine blood vessels.
步骤S22、建立血管区域特征函数：基于血管特性并根据所述特征值和特征向量建立血管区域特征函数。Step S22 , establishing the characteristic function of the blood vessel area: establishing the characteristic function of the blood vessel area based on the characteristics of the blood vessel and according to the eigenvalues and the eigenvectors.
现有的血管区域增强算法：根据所述特征值λ1、λ2、λ3和所述特征向量γ1、γ2、γ3，首先定义了变量RA、RB及S，分别如下式(4)、(5)、(6)：Existing blood vessel region enhancement algorithm: According to the eigenvalues λ1 , λ2 , λ3 and the eigenvectors γ1 , γ2 , γ3 , the variablesRA ,RB and S are first defined, respectively as follows: (4), (5), (6):
RA＝|λ2|/|λ3| (4)；RA = |λ2 |/|λ3 | (4);
变量RA的值可以估计当前处理图像属于管状结构的可能性，其值越大，越可能属于管状结构；The value of the variable RA can estimate the possibility that the currently processed image belongs to the tubular structure, and the larger the value, the more likely it belongs to the tubular structure;
变量RB的值可以估计当前处理图像属于盘状结构的可能性，其值越大，越可能属于盘状结构；The value of the variableRB can estimate the possibility that the currently processed image belongs to the discoid structure, and the larger the value, the more likely it belongs to the discoid structure;
变量S取所述三个特征值的平方和，可以模糊判断当前处理图像属于结构性较强的区域。The variable S takes the sum of the squares of the three eigenvalues, which can be used to fuzzily determine that the currently processed image belongs to an area with strong structure.
根据所述变量RA、RB及S定义的现有血管区域函数为下式(7)：The existing blood vessel area function defined according to the variablesRA ,RB and S is the following formula (7):
其中，a、b以及c分别为平面结构参数、球状结构参数及背景区分参数，根据经验以及所选尺度选取。Among them, a, b, and c are planar structure parameters, spherical structure parameters, and background discrimination parameters, respectively, which are selected according to experience and selected scales.
针对真实的三维精细血管图像，血管粗细及拓扑都不规则，血管粗细不等，对于非常细的血管，其交叉点几乎可以忽略，但是对于较粗的血管，其分叉区域就不容忽视了。For real three-dimensional fine blood vessel images, the thickness and topology of blood vessels are irregular, and the thickness of blood vessels varies. For very thin blood vessels, the intersection points can be almost ignored, but for thicker blood vessels, the bifurcation area cannot be ignored.
根据公式(5)计算变量RB时，若当前处理图像为盘状结构或分叉区域，λ1是和λ2、λ3大小相当的负数，这两种情况下，变量RB的值均较大；若当前处理图像为血管弧度较大的区域，RB的值也会较大。When the variable RB is calculated according to formula (5), if the currently processed image is a disk-shaped structure or a bifurcation area, λ1 is a negative number comparable to λ2 and λ3 , in both cases, the value of the variableRB is equal to Larger; if the currently processed image is an area with a large vascular arc, the value ofRB will also be larger.
但根据公式(7)计算现有血管区域函数时，RB越小，当前处理图像属于血管区域的权值越大，因而在特殊情况下，根据公式(7)判断当前处理图像会出现错误，所以应该找回被忽略的分叉区域和血管弧度较大的区域。However, when calculating the function of the existing blood vessel area according to formula (7), the smaller theRB , the greater the weight of the current processing image belonging to the blood vessel area. Therefore, in special cases, it will be wrong to judge the current processing image according to formula (7). Therefore, the neglected bifurcation area and the area with large vascular curvature should be retrieved.
本发明假设所述特征向量γ1、γ2、γ3的反方向分别对应为τ1,τ2,τ3，所述特征向量γ1、γ2、γ3指向血管壁，在一定概率下，τ1,τ2,τ3中有至少一个方向指向血管延伸方向，如果沿着τ1,τ2,τ3其中某一方向像素灰度值变化缓慢，则可以判定为有效的血管区域。The present invention assumes that the opposite directions of the eigenvectors γ1 , γ2 , and γ3 correspond to τ1 , τ2 , and τ3 respectively, and the eigenvectors γ1 , γ2 , and γ3 point to the vessel wall. Under a certain probability , τ1 , τ2 , τ3 have at least one direction pointing to the blood vessel extension direction, and if the pixel gray value changes slowly along one of τ1 , τ2 , τ3 directions, it can be determined as an effective blood vessel region.
定义沿方向τ的梯度方向对于尺度σ，计算得到沿此梯度方向的变化率的大小k，设三个方向τ1,τ2,τ3的灰度变化率大小分别为k1,k2,k3，本发明定义新变量即血管角点区域特性RD，如下式(8)：Define the direction of the gradient along direction τ For the scale σ, the magnitude k of the rate of change along the gradient direction is calculated, and the grayscale change rates of the three directions τ1 , τ2 , τ3 are respectively k1 , k2 , k3 , and the present invention defines a new The variable is the characteristic RD of the blood vessel corner area, as shown in the following formula (8):
根据所述变量RA、RB及新变量RD，定义新的血管区域函数为下式(9)：According to the variablesRA ,RB and the new variableRD , the new blood vessel area function is defined as the following formula (9):
其中，a、c以及d分别为平面结构参数、背景区分参数以及三维血管结构参数，根据经验以及所选尺度选取。Wherein, a, c, and d are planar structure parameters, background discrimination parameters, and three-dimensional blood vessel structure parameters, selected according to experience and selected scales.
若当前处理图像为分叉区域，k1、k2及k3中只要任意一个值较小，则变量RD的值较大；若当前处理图像为血管弧度较大的区域，k2及k3与λ2及λ3大小相当，k1的值相对较小，则变量RD的值较大；若当前处理图像为典型的直血管区域，k1≈0，变量RD的值依然较大；因此新变量RD适用多种血管区域，不会对其他特征项产生干扰。If the current processing image is a bifurcation area, as long as any value of k1 , k2 and k3 is small, the value of the variable RD is large; if the current processing image is a region with a large vascular arc, k2 and k3 is equivalent to λ2 and λ3 , the value of k1 is relatively small, and the value of variableRD is relatively large; if the current processing image is a typical straight blood vessel area, k1 ≈ 0, the value of variableRD is still relatively small. Large; therefore, the new variable RD is applicable to a variety of vascular regions and will not interfere with other feature items.
步骤S23、确定需要增强的血管区域：根据所述血管区域特征函数确定需要增强的血管区域，进而对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像。所述三维精细血管增强图像包括增强后的待修复图像1和增强后的参考图像3。Step S23 , determining the vessel region to be enhanced: determining the vessel region to be enhanced according to the characteristic function of the vessel region, and then enhancing the original 3D fine blood vessel image to obtain a 3D fine blood vessel enhanced image. The three-dimensional fine blood vessel enhanced image includes an enhanced image to be repaired 1 and an enhanced reference image 3 .
在本实施例中，具体地，请参见图6(a)及图6(b)，其中，图6(a)为图3(a)所示增强后的参考图像；图6(b)为图3(b)所示增强后的待修复图像。对比发现利用本发明提供的血管区域特征函数进行图像增强具有较好的效果。In this embodiment, specifically, please refer to FIG. 6(a) and FIG. 6(b), wherein, FIG. 6(a) is the enhanced reference image shown in FIG. 3(a); FIG. 6(b) is Figure 3(b) shows the enhanced image to be repaired. By comparison, it is found that image enhancement using the blood vessel region feature function provided by the present invention has a better effect.
请参阅图7，为图2所示三维精细血管重建方法中步骤S3的流程图。所述基于血管特性的特征匹配步骤S3包括如下步骤：Please refer to FIG. 7 , which is a flowchart of step S3 in the three-dimensional fine vessel reconstruction method shown in FIG. 2 . The feature matching step S3 based on blood vessel characteristics includes the following steps:
步骤S31、将所述待修复图像1均分为多个与所述参考映射块32大小相同的候选块，计算全局匹配度SM。Step S31, divide the image to be repaired 1 into multiple candidate blocks with the same size as the reference mapping block 32, and calculate the global matching degree SM .
根据所述三维精细血管增强图像，即所述待修复图像1和所述参考图像3，对两幅图像中血管的整体分布进行相似度匹配，计算所述候选块和所述参考映射块32的全局匹配度SM。According to the three-dimensional fine blood vessel enhanced image, that is, the image to be repaired 1 and the reference image 3, perform similarity matching on the overall distribution of blood vessels in the two images, and calculate the candidate block and the reference mapping block 32 Global matching degree SM .
新建0-1矩阵B1，将所述待修复图像1的所述候选块中属于血管的相应位置设为1，属于背景的相应位置设为0；新建0-1矩阵B2，将所述参考图像3的所述参考映射块32中属于血管的相应位置设为1，属于背景的相应位置设为0；对所述矩阵B1和所述矩阵B2相应位置上进行“异或运算”，得到结果矩阵MR，所述矩阵MR中“0”越多，即相同之处越多，则说明所述待修复图像1的所述候选块和所述参考图像3的所述参考映射块32从轮廓上看越相似。Create a new 0-1 matrix B1 , set the corresponding position belonging to the blood vessel in the candidate block of the image 1 to be repaired to 1, and set the corresponding position belonging to the background to 0; create a new 0-1 matrix B2 , set the In the reference mapping block 32 of the reference image 3, the corresponding position belonging to the blood vessel is set to 1, and the corresponding position belonging to the background is set to 0; an "exclusive OR operation" is performed on the corresponding positions of the matrix B1 and the matrix B2 , to obtain the result matrix MR , the more "0"s in the matrix MR , that is, the more similarities, it means that the candidate block of the image to be repaired 1 and the reference mapping of the reference image 3 Blocks 32 are more similar in outline.
定义全局匹配度SM为下式(10)：Define the global matching degree SM as the following formula (10):
其中，定义u(x,y,z)为下式(11)：Among them, u(x,y,z) is defined as the following formula (11):
步骤S32、计算血管段相似度ST。Step S32, calculating the blood vessel segment similarity ST .
请参阅图8，为图7所示步骤S3中步骤S32的流程图。所述计算血管段相似度ST的步骤S32包括：Please refer to FIG. 8 , which is a flowchart of step S32 in step S3 shown in FIG. 7 . The step S32 of calculating the similarityST of the blood vessel segment includes:
步骤S32-1、基于逐层剥取法细化步骤S2中所述需要增强的血管区域，提取单像素中心线，由所述三维精细血管增强图像得到三维精细血管二值图，所述三维精细血管二值图包括所述待修复图像1的二值图和所述参考图像3的二值图；Step S32-1. Based on the layer-by-layer stripping method, refine the blood vessel region to be enhanced in step S2, extract the single-pixel centerline, and obtain a three-dimensional fine blood vessel binary image from the three-dimensional fine blood vessel enhanced image, and the three-dimensional fine blood vessel The binary image includes the binary image of the image to be repaired 1 and the binary image of the reference image 3;
步骤S32-2、根据所述待修复图像1的二值图提取所述候选块的端点集，根据所述参考图像3的二值图提取所述参考映射块32的端点集，分别求取端点特征向量；Step S32-2, extract the endpoint set of the candidate block according to the binary image of the image to be repaired 1, extract the endpoint set of the reference mapping block 32 according to the binary image of the reference image 3, and obtain the endpoints respectively Feature vector;
步骤S32-3、根据所述端点特征向量计算血管段相似度ST。Step S32-3, calculating the blood vessel segment similarity ST according to the endpoint feature vector.
具体过程为：The specific process is:
对于血管图像，血管的分段、分叉点以及端点是特别需要关注的特征，因此利用骨架化思想，提取步骤S2中所述需要增强的血管区域的单像素宽的中心线，由所述三维精细血管增强图像得到了三维精细血管二值图，所述三维精细血管二值图包括所述待修复图像1的二值图和所述参考图像3的二值图；具体请参见图9(a)及图9(b)，其中图9(a)为图6(a)所示细化后的参考图像；图9(b)为图6(b)所示细化后的待修复图像。For blood vessel images, the segmentation, bifurcation points, and endpoints of blood vessels are the features that need special attention. Therefore, using the skeletonization idea, extract the single-pixel-wide centerline of the blood vessel area that needs to be enhanced in step S2, and use the three-dimensional The fine blood vessel enhanced image obtains a three-dimensional fine blood vessel binary image, and the three-dimensional fine blood vessel binary image includes the binary image of the image to be repaired 1 and the binary image of the reference image 3; for details, please refer to FIG. 9(a ) and Figure 9(b), wherein Figure 9(a) is the reference image after refinement shown in Figure 6(a); Figure 9(b) is the image to be repaired after refinement shown in Figure 6(b).
细化后的单像素图像由一段段弯曲的线段组成，其中端点可分为终点和交叉点；有些端点从本身分开两条或两条以上的线，这些端点就是交叉点，而有些端点在它本身处终止，这些端点是终点。The thinned single-pixel image is composed of curved line segments, and the endpoints can be divided into endpoints and intersections; some endpoints separate two or more lines from themselves, these endpoints are intersections, and some endpoints are in it terminates at itself, these endpoints are endpoints.
设所述待修复图像1中所述候选块的二值图为体数据V1，提取所述候选块的端点集，所述端点集包括终点集ES1＝{E11,E12,…,E1m}和分叉点集FS1＝{F11,F12,…,F1n}，设所述体数据V1的边长为a，则将其划分为8×8×8个小的小立方块(边长为a/8)，统计落在各个小立方块中的终点数量e1n(n＝1,2,…512)和分叉点数量f1n(n＝1,2,…512)，组成一个带有位置信息的所述体数据V1的端点特征向量，所述端点特征向量包括终点特征向量和分叉点特征向量设所述参考图像3中所述参考映射块32的二值图为体数据V2，相同方法获得所述体数据V2的端点特征向量，所述端点特征向量包括终点特征向量和分叉点特征向量Let the binary image of the candidate block in the image to be repaired 1 be the volume data V1 , extract the endpoint set of the candidate block, the endpoint set includes the endpoint set ES1 ={E11 ,E12 ,..., E1m } and bifurcation point set FS1 ={F11 ,F12 ,…,F1n }, assuming that the side length of the volume data V1 is a, it is divided into 8×8×8 small Small cubes (side length is a/8), count the number of end points e1n (n=1,2,...512) and the number of bifurcation points f1n (n=1,2,...512) falling in each small cube 512), forming an endpoint feature vector of the volume data V1 with position information, the endpoint feature vector including the endpoint feature vector and bifurcation point eigenvectors Assuming that the binary image of the reference mapping block 32 in the reference image 3 is volume data V2 , the endpoint feature vector of the volume data V2 is obtained in the same way, and the endpoint feature vector includes the endpoint feature vector and bifurcation point eigenvectors
利用余弦定理求取两组体数据的终点特征向量和的相关度STe，如下式(12)；以及分叉点特征向量和的相关度STf，如下式(13)；根据所述相关度STe和所述相关度STf计算所述血管段相似度ST，如下式(14)：Obtaining the Endpoint Eigenvectors of Two Groups of Volume Data Using the Law of Cosines and The correlation STe , as shown in the following formula (12); and the eigenvector of the bifurcation point and The degree of correlation STf is shown in the following formula (13); the blood vessel segment similarity ST is calculated according to the degree of correlation STe and the degree of correlation STf is shown in the following formula (14):
ST＝STe·STf (14)。ST = STe · STf (14).
步骤S33、根据所述全局匹配度SM和所述血管段相似度ST计算整体匹配度Sves，所述整体匹配度Sves＝μSM+ηST，其中，μ及η分别为SM及ST的权重系数，且μ+η＝1。Step S33, calculating the overall matching degree Sves according to the global matching degree SM and the blood vessel segment similarity ST , the overall matching degree Sves =μSM +ηST , where μ and η are respectively SM And the weight coefficient ofST , and μ+η=1.
步骤S34、根据所述整体匹配度Sves，得到所述参考映射块32在所述待修复图像1中的最优匹配块52，所述参考映射块32中的参考映射区域30对应所述最优匹配块52中的填充区域50。Step S34. According to the overall matching degree Sves , the optimal matching block 52 of the reference mapping block 32 in the image to be repaired 1 is obtained, and the reference mapping area 30 in the reference mapping block 32 corresponds to the optimal matching block 52. Padding area 50 in block 52 is best matched.
所述最优匹配块52为整体匹配度最高的所述候选块。所述最优匹配块52通过映射位置关系获得与所述待修复区域10相对应的填充区域50。The optimal matching block 52 is the candidate block with the highest overall matching degree. The optimal matching block 52 obtains the filling area 50 corresponding to the area to be repaired 10 by mapping the positional relationship.
请参阅图10，为图2所示三维精细血管重建方法中步骤S4的流程图。所述图像填充及边界处理步骤S4包括：Please refer to FIG. 10 , which is a flowchart of step S4 in the three-dimensional fine vessel reconstruction method shown in FIG. 2 . The image filling and boundary processing step S4 includes:
步骤S41、将所述最优匹配块52的填充区域50对应填充至所述待修复块12的待修复区域10；Step S41, correspondingly fill the filling area 50 of the optimal matching block 52 into the area 10 to be repaired of the block 12 to be repaired;
步骤S42、将所述填充区域50的边界进行插值迭代，不断逼近直至收敛，将所述填充区域50、以及所述待修复区域10的边界平滑过渡，完成修复重建。Step S42 , interpolating and iterating the boundary of the filling area 50 until it converges, making a smooth transition between the filling area 50 and the boundary of the area to be repaired 10 , and completing the restoration and reconstruction.
具体过程为：The specific process is:
将所述填充区域50填充到所述待修复区域10，所述填充区域50覆盖所述待修复区域10完成所述缺失部位的修复。The filling area 50 is filled into the area to be repaired 10 , and the filling area 50 covers the area to be repaired 10 to complete the repair of the missing part.
采用“逼近”的思想将所述待修复区域10的边界与所述填充区域50的像素灰度平滑过渡，弱化其之间拼缝的边缘。对所述待修复块12与所述最优匹配块52的限定范围分别建立矢量场，在矢量场的引导下来进行插值迭代。The idea of "approximation" is adopted to make a smooth transition between the boundary of the region to be repaired 10 and the pixel grayscale of the filling region 50, and weaken the edge of the patchwork between them. Vector fields are respectively established for the limited ranges of the block to be repaired 12 and the optimal matching block 52 , and interpolation iterations are performed under the guidance of the vector fields.
设点p(x,y,z)是所述待修复块12中一点，其6个邻域点的集合为N6，定义其6个邻域方向的差分矢量集合为6个邻域方向上的单位矢量分别为根据差分的定义得到Let point p(x, y, z) be a point in the block to be repaired 12, the set of its 6 neighborhood points is N6 , and the set of difference vectors defining its 6 neighborhood directions is The unit vectors in the six neighborhood directions are According to the definition of difference
对所述待修复块12中每个像素点都进行上述操作，得到所述待修复块12中所有像素点所对应集合所有得到的矢量组成了所述待修复块12的矢量场D1，进行数据迭代，最终得到像素集合P。Perform the above operations on each pixel in the block to be repaired 12 to obtain a set corresponding to all pixels in the block to be repaired 12 All obtained vectors constitute the vector field D1 of the block to be repaired 12 , and data iteration is performed to finally obtain a pixel set P.
相同方法构造所述最优匹配块52的矢量场D5，进行数据迭代，最终得到像素集合Q。In the same way, the vector field D5 of the optimal matching block 52 is constructed, data iteration is performed, and the pixel set Q is finally obtained.
为了使所述像素集合P逼近所述最像素集合Q，该问题等价于下式(15)的最小值问题：In order to make the pixel set P approach the most pixel set Q, this problem is equivalent to the minimum value problem of the following formula (15):
其中，Δ是某像素点六邻域方向上的差分算子，D1是所述待修复块12的矢量场。根据Euler-Lagrange方程，上述问题等价于下式(16)：Wherein, Δ is a difference operator in the six-neighborhood direction of a certain pixel point, and D1 is the vector field of the block 12 to be repaired. According to the Euler-Lagrange equation, the above problem is equivalent to the following formula (16):
Δp＝D1,p∈D1 (16)；Δp=D1 , p∈D1 (16);
其意义为以矢量场为参照来实现像素的无限逼近。下面采用超松弛迭代法，通过邻域像素消元来实现。Its meaning is to realize the infinite approximation of pixels with the vector field as a reference. Next, the super-relaxation iterative method is used to realize it through neighborhood pixel elimination.
设初始化图像像素函数值为p0＝p(x,y,z),(x,y,z)∈D1，迭代的过程为下式(17)：Assuming that the initial image pixel function value is p0 =p(x,y,z),(x,y,z)∈D1 , the iterative process is the following formula (17):
其中，Δpt(x,y,z)如下式(18)：Among them, Δpt (x, y, z) is as follows (18):
Δpt(x,y,z)＝pt(x+1,y,z)+pt(x-1,y,z)+pt(x,y+1,z)+pt(x,y-1,z)+pt(x,y,z+1)+pt(x,y,z-1)-6×pt(x,y,z) (18)；Δpt (x, y, z) = pt (x+1, y, z) + pt (x-1, y, z) + pt (x, y+1, z) + pt (x ,y-1,z)+pt (x,y,z+1)+pt (x,y,z-1)-6×pt (x,y,z) (18);
每一次迭代，都会更新所述待修复块12的像素值，使之在矢量场引导下逐渐逼近所述最优匹配块52内相应位置的像素值，直到收敛即可实现灰度协调过渡的效果。Each iteration, the pixel value of the block to be repaired 12 will be updated, so that it will gradually approach the pixel value of the corresponding position in the optimal matching block 52 under the guidance of the vector field, until the convergence can achieve the effect of gray scale coordination transition .
应用所述三维精细血管重建方法对三维精细血管进行修复，待修复的三维精细血管结构图和修复后的三维精细血管结构图详见图11(a)及图11(b)，可以得知，本发明应用该方法能够很好的修复具缺失部位的三维精细血管，精度更高。The three-dimensional fine blood vessel reconstruction method is used to repair the three-dimensional fine blood vessel. The three-dimensional fine blood vessel structure diagram to be repaired and the repaired three-dimensional fine blood vessel structure diagram are shown in Figure 11(a) and Figure 11(b). It can be known that, By applying the method in the present invention, the three-dimensional fine blood vessels with missing parts can be well repaired, and the precision is higher.
请参见图12，为本发明提供的三维精细血管重建系统的结构框图。所述三维精细血管重建系统100包括图像载入模块11、图像预处理模块13、特征匹配模块15及填充处理模块17。Please refer to FIG. 12 , which is a structural block diagram of the three-dimensional fine vessel reconstruction system provided by the present invention. The three-dimensional fine vessel reconstruction system 100 includes an image loading module 11 , an image preprocessing module 13 , a feature matching module 15 and a filling processing module 17 .
所述图像载入模块11，用于载入三维精细血管原始图像，所述三维精细血管原始图像包括扫描同一对象的待修复图像1和参考图像3，所述待修复图像1为具缺失的血管图像，所述参考图像3为完整的血管图像，所述待修复图像1的精度高于所述参考图像3；并根据所述待修复图像1指定待修复区域10，并定义包含所述待修复区域10的待修复块12，根据所述待修复块12确定在所述参考图像3中位置相应、大小相同的参考映射块32，所述待修复块12中的待修复区域10对应所述参考映射块32中的参考映射区域30。所述图像载入模块11的详细执行流程对应如上步骤S1所述，不再赘述。The image loading module 11 is configured to load an original image of a three-dimensional fine blood vessel, the original image of a three-dimensional fine blood vessel includes an image to be repaired 1 and a reference image 3 that scan the same object, and the image to be repaired 1 is a missing blood vessel image, the reference image 3 is a complete blood vessel image, and the accuracy of the image to be repaired 1 is higher than that of the reference image 3; and according to the image to be repaired 1, the region to be repaired 10 is specified, and the area to be repaired is defined to include the The block 12 to be repaired in the area 10, according to the block 12 to be repaired, the reference mapping block 32 with the corresponding position and the same size in the reference image 3 is determined, and the area 10 to be repaired in the block 12 to be repaired corresponds to the reference The reference map region 30 in the map block 32 . The detailed execution flow of the image loading module 11 corresponds to that described in step S1 above, and will not be repeated here.
所述图像预处理模块13，用于对所述图像载入模块11载入的所述三维精细血管原始图像进行分析，确定需要增强的血管区域，对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像，所述三维精细血管增强图像包括增强后的所述待修复图像1和增强后的所述参考图像3。The image preprocessing module 13 is configured to analyze the original three-dimensional fine blood vessel image loaded by the image loading module 11, determine the blood vessel region to be enhanced, and enhance the original three-dimensional fine blood vessel image to obtain A three-dimensional fine blood vessel enhanced image, the three-dimensional fine blood vessel enhanced image includes the enhanced image 1 to be repaired and the enhanced reference image 3 .
所述特征匹配模块15，用于根据所述图像预处理模块13得到的三维精细血管增强图像，计算整体匹配度，得到所述参考映射块32在所述待修复图像1中的最优匹配块52，所述参考映射块32中的参考映射区域30对应所述最优匹配块52中的填充区域50。The feature matching module 15 is configured to calculate the overall matching degree according to the three-dimensional fine blood vessel enhanced image obtained by the image preprocessing module 13, and obtain the optimal matching block of the reference mapping block 32 in the image to be repaired 1 52 , the reference mapping area 30 in the reference mapping block 32 corresponds to the padding area 50 in the optimal matching block 52 .
所述填充处理模块17，用于将所述特征匹配模块15得到的所述最优匹配块52的填充区域50对应填充至所述待修复块12的待修复区域10，并对边界进行平滑处理，完成修复重建。The filling processing module 17 is configured to fill the filling area 50 of the optimal matching block 52 obtained by the feature matching module 15 into the area to be repaired 10 of the block to be repaired 12, and smooth the boundary , to complete the restoration and reconstruction.
请参照图13，为图12所示图像预处理模块的结构框图。所述图像预处理模块13包括第一运算单元131、函数创建单元133及血管增强单元135。Please refer to FIG. 13 , which is a structural block diagram of the image preprocessing module shown in FIG. 12 . The image preprocessing module 13 includes a first calculation unit 131 , a function creation unit 133 and a blood vessel enhancement unit 135 .
所述第一运算单元131，选取Hessian矩阵对所述三维精细血管原始图像做卷积运算，求得特征值和特征向量。The first computing unit 131 selects a Hessian matrix to perform a convolution operation on the original three-dimensional fine blood vessel image to obtain eigenvalues and eigenvectors.
所述函数创建单元133，根据所述第一运算单元131求得的特征值和特征向量并基于血管特性建立血管区域特征函数。The function creation unit 133 establishes a blood vessel region feature function based on the vascular characteristics based on the eigenvalues and eigenvectors obtained by the first computing unit 131 .
所述血管增强单元135，根据所述函数创建单元133建立的血管区域特征函数确定需要增强的血管区域，进而对所述三维精细血管原始图像进行增强，得到三维精细血管增强图像。其中，每一单元的详细执行流程对应如上步骤S21到S23所述，不再赘述。The blood vessel enhancement unit 135 determines the blood vessel region to be enhanced according to the blood vessel region feature function established by the function creation unit 133, and then enhances the original 3D fine blood vessel image to obtain a 3D fine blood vessel enhanced image. Wherein, the detailed execution flow of each unit corresponds to the above steps S21 to S23, and will not be repeated here.
请参照图14，为图12所示特征匹配模块的结构框图。所述特征匹配模块15包括第一计算单元151、第二计算单元152及第三计算单元153，分别用于计算全局匹配度SM、血管段相似度ST及整体匹配度Sves。其中，每一单元的详细执行流程对应如上步骤S31到S34所述，不再赘述。Please refer to FIG. 14 , which is a structural block diagram of the feature matching module shown in FIG. 12 . The feature matching module 15 includes a first calculation unit 151 , a second calculation unit 152 and a third calculation unit 153 , which are respectively used to calculate the global matching degree SM , the vessel segment similarity ST and the overall matching degreeSves . Wherein, the detailed execution flow of each unit corresponds to the above steps S31 to S34, and will not be repeated here.
请参照图15，为图12所示填充处理模块的结构框图。所述填充处理模块17包括图像填充单元171及边界处理单元173。Please refer to FIG. 15 , which is a structural block diagram of the filling processing module shown in FIG. 12 . The filling processing module 17 includes an image filling unit 171 and a boundary processing unit 173 .
所述图像填充单元171，将所述最优匹配块52的填充区域50对应填充至所述待修复块12的待修复区域10。The image filling unit 171 correspondingly fills the filling area 50 of the optimal matching block 52 into the area 10 of the block to be repairing 12 to be repaired.
所述边界处理单元173，将所述填充区域50的边界进行插值迭代，不断逼近直至收敛，将所述填充区域50、以及所述待修复区域10的边界平滑过渡，完成修复重建。其中，每一单元的详细执行流程对应如上步骤S41到S42所述，不再赘述。The boundary processing unit 173 performs interpolation iteration on the boundary of the filling area 50 until it converges, and smoothly transitions the boundary between the filling area 50 and the area to be repaired 10 to complete the restoration and reconstruction. Wherein, the detailed execution flow of each unit corresponds to the above steps S41 to S42, and will not be repeated here.
本发明提供的三维精细血管重建方法及其系统具有以下有益效果：The three-dimensional fine vessel reconstruction method and system thereof provided by the present invention have the following beneficial effects:
一、本发明提供的所述三维精细血管重建方法对需要增强的血管区域进行了特征值提取，将所述三维精细血管原始图像进行了图像增强，便于对源于同一血管样本的所述待修复图像1和所述参考图像3进行结构匹配，提高了三维精细血管重建的效率，并能够保证精度；1. The three-dimensional fine blood vessel reconstruction method provided by the present invention extracts the feature value of the blood vessel region to be enhanced, and performs image enhancement on the original image of the three-dimensional fine blood vessel, so as to facilitate the reconstruction of the to-be-repaired blood vessel from the same blood vessel sample. Structural matching of the image 1 and the reference image 3 improves the efficiency of three-dimensional fine vessel reconstruction and ensures accuracy;
二、本发明重新定义了基于Hessian阵的所述血管区域特征函数，考虑了所述三维精细血管原始图像的灰度信息，可以有效提高图像增强的效果；同时考虑了血管的分叉区域以及弧度较大的区域等个体差异性的影响，能够很好的解决个体差异性的问题，即使在特殊情况也不会判断错误，具有更广泛的适用性；2. The present invention redefines the characteristic function of the blood vessel region based on the Hessian matrix, and considers the gray information of the original image of the three-dimensional fine blood vessel, which can effectively improve the effect of image enhancement; at the same time, the bifurcation area and radian of the blood vessel are considered The influence of individual differences such as large areas can well solve the problem of individual differences, and it will not make mistakes in judgment even in special cases, and has wider applicability;
三、本发明提供的所述三维精细血管重建方法基于血管特性而成，考虑到血管结构的管状特性和较强的自相似性，从而指导源于同一血管样本的所述待修复图像1和所述参考图像3的匹配，提高了血管结构特征匹配的可靠性和准确性，进一步增大三维精细血管重建的精度；3. The three-dimensional fine blood vessel reconstruction method provided by the present invention is based on the characteristics of blood vessels, taking into account the tubular characteristics and strong self-similarity of the blood vessel structure, so as to guide the image 1 to be repaired and the image to be repaired from the same blood vessel sample The matching of the above-mentioned reference image 3 improves the reliability and accuracy of vascular structure feature matching, and further increases the precision of three-dimensional fine vessel reconstruction;
四、本发明通过所述待修复图像1和所述参考图像3的匹配，得到所述参考映射块32在所述待修复图像1中的所述最优匹配块52，对应填充之后再对边界进行平滑处理，修复精度高。4. The present invention obtains the optimal matching block 52 of the reference mapping block 32 in the image to be repaired 1 through the matching of the image to be repaired 1 and the reference image 3, and then corrects the boundary after corresponding padding Perform smoothing and repair with high precision.
所属技术领域的技术人员可以理解：实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成，前述的程序可以存储于计算机可读取存储介质中，并被处理器执行，前述的程序在被执行时处理器可以执行包括上述方法实施例的全部或者部分步骤。其中，所述处理器可以作为一个或多个处理器芯片实施，或者可以为一个或多个专用集成电路(Application Specific Integrated Circuit,ASIC)的一部分；而前述的存储介质可以包括但不限于以下类型的存储介质：闪存(Flash Memory)、只读存储器(Read-OnlyMemory,ROM)、随机存取存储器(Random Access Memory,RAM)、移动硬盘、磁碟或者光盘等各种可以存储程序代码的介质。Those skilled in the art can understand that all or part of the steps for implementing the above method embodiments can be completed by hardware related to program instructions. The aforementioned programs can be stored in computer-readable storage media and executed by a processor. The aforementioned When the program is executed, the processor may perform all or part of the steps including the above method embodiments. Wherein, the processor can be implemented as one or more processor chips, or can be a part of one or more application specific integrated circuits (Application Specific Integrated Circuit, ASIC); and the aforementioned storage medium can include but not limited to the following types Storage media: flash memory (Flash Memory), read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), mobile hard disk, magnetic disk or optical disk and other media that can store program code.
以上所述仅为本发明的优选实施例而已，并不用于限制本发明，对于本领域的技术人员来说，本发明可以有各种更改和变化。凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610305637.0ACN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610305637.0ACN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
| Publication Number | Publication Date |
|---|---|
| CN106127849Atrue CN106127849A (en) | 2016-11-16 |
| CN106127849B CN106127849B (en) | 2019-01-11 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610305637.0AExpired - Fee RelatedCN106127849B (en) | 2016-05-10 | 2016-05-10 | Three-dimensional fine vascular method for reconstructing and its system |
| Country | Link |
|---|---|
| CN (1) | CN106127849B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106875375A (en)* | 2016-12-28 | 2017-06-20 | 浙江工业大学 | Three-dimensional blood vessel axis detection method based on tubulose feature enhancing filtering and ridge line tracking |
| CN107067398A (en)* | 2017-04-28 | 2017-08-18 | 青岛海信医疗设备股份有限公司 | Complementing method and device for lacking blood vessel in 3 D medical model |
| CN109242964A (en)* | 2018-11-01 | 2019-01-18 | 青岛海信医疗设备股份有限公司 | The treating method and apparatus of 3 D medical model |
| CN109829489A (en)* | 2019-01-18 | 2019-05-31 | 刘凯欣 | A kind of cultural relic fragments recombination method and device based on multilayer feature |
| CN110120027A (en)* | 2019-04-16 | 2019-08-13 | 华南理工大学 | The CT sectioning image Enhancement Method and device of Machine oriented learning system data |
| CN111968051A (en)* | 2020-08-10 | 2020-11-20 | 珠海普生医疗科技有限公司 | Endoscope blood vessel enhancement method based on curvature analysis |
| CN112330708A (en)* | 2020-11-24 | 2021-02-05 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN112700420A (en)* | 2020-12-31 | 2021-04-23 | 中南大学 | Eye fundus image complementing and classifying method and system |
| CN112837288A (en)* | 2021-02-01 | 2021-05-25 | 数坤（北京）网络科技有限公司 | Blood vessel centerline extraction method and device and readable storage medium |
| KR20230021544A (en)* | 2021-08-05 | 2023-02-14 | 숭실대학교산학협력단 | Method for coronary arteries non-rigid registration using hierarchical deformation in computed tomography angiography images recording medium and device for performing the method |
| CN116596950A (en)* | 2023-05-31 | 2023-08-15 | 东北林业大学 | Retina fundus blood vessel tracking method based on feature weighted clustering |
| CN116721760A (en)* | 2023-06-12 | 2023-09-08 | 东北林业大学 | Biomarker-fused multitasking diabetic retinopathy detection algorithm |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130027389A1 (en)* | 2011-07-27 | 2013-01-31 | International Business Machines Corporation | Making a two-dimensional image into three dimensions |
| CN102999887A (en)* | 2012-11-12 | 2013-03-27 | 中国科学院研究生院 | Sample based image repairing method |
| CN104361626A (en)* | 2014-09-29 | 2015-02-18 | 北京理工大学 | Subcutaneous vein three-dimensional reconstruction method based on hybrid matching strategy |
| CN104933756A (en)* | 2014-03-21 | 2015-09-23 | 北京冠生云医疗技术有限公司 | Construction method of three-dimensional coronary artery analysis model and system thereof |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130027389A1 (en)* | 2011-07-27 | 2013-01-31 | International Business Machines Corporation | Making a two-dimensional image into three dimensions |
| CN102999887A (en)* | 2012-11-12 | 2013-03-27 | 中国科学院研究生院 | Sample based image repairing method |
| CN104933756A (en)* | 2014-03-21 | 2015-09-23 | 北京冠生云医疗技术有限公司 | Construction method of three-dimensional coronary artery analysis model and system thereof |
| CN104361626A (en)* | 2014-09-29 | 2015-02-18 | 北京理工大学 | Subcutaneous vein three-dimensional reconstruction method based on hybrid matching strategy |
| Title |
|---|
| 王敏敏: ""基于紧小波框架的多幅盲运动图像修复算法研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》* |
| 顾晨: ""增强现实中的图像修复算法"", 《中国优秀硕士学位论文全文数据库 信息科技辑》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106875375B (en)* | 2016-12-28 | 2019-07-30 | 浙江工业大学 | A kind of three-dimensional blood vessel axis detection method based on tubulose signature tracking |
| CN106875375A (en)* | 2016-12-28 | 2017-06-20 | 浙江工业大学 | Three-dimensional blood vessel axis detection method based on tubulose feature enhancing filtering and ridge line tracking |
| CN107067398A (en)* | 2017-04-28 | 2017-08-18 | 青岛海信医疗设备股份有限公司 | Complementing method and device for lacking blood vessel in 3 D medical model |
| CN109242964B (en)* | 2018-11-01 | 2023-04-11 | 青岛海信医疗设备股份有限公司 | Processing method and device of three-dimensional medical model |
| CN109242964A (en)* | 2018-11-01 | 2019-01-18 | 青岛海信医疗设备股份有限公司 | The treating method and apparatus of 3 D medical model |
| CN109829489A (en)* | 2019-01-18 | 2019-05-31 | 刘凯欣 | A kind of cultural relic fragments recombination method and device based on multilayer feature |
| CN110120027A (en)* | 2019-04-16 | 2019-08-13 | 华南理工大学 | The CT sectioning image Enhancement Method and device of Machine oriented learning system data |
| CN110120027B (en)* | 2019-04-16 | 2021-05-14 | 华南理工大学 | CT slice image enhancement method and device for machine learning system data |
| CN111968051A (en)* | 2020-08-10 | 2020-11-20 | 珠海普生医疗科技有限公司 | Endoscope blood vessel enhancement method based on curvature analysis |
| CN112330708A (en)* | 2020-11-24 | 2021-02-05 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
| CN112330708B (en)* | 2020-11-24 | 2024-04-23 | 沈阳东软智能医疗科技研究院有限公司 | Image processing method, device, storage medium and electronic equipment |
| CN112700420A (en)* | 2020-12-31 | 2021-04-23 | 中南大学 | Eye fundus image complementing and classifying method and system |
| CN112837288A (en)* | 2021-02-01 | 2021-05-25 | 数坤（北京）网络科技有限公司 | Blood vessel centerline extraction method and device and readable storage medium |
| KR102629331B1 (en) | 2021-08-05 | 2024-01-29 | 숭실대학교 산학협력단 | Method for coronary arteries non-rigid registration using hierarchical deformation in computed tomography angiography images recording medium and device for performing the method |
| KR20230021544A (en)* | 2021-08-05 | 2023-02-14 | 숭실대학교산학협력단 | Method for coronary arteries non-rigid registration using hierarchical deformation in computed tomography angiography images recording medium and device for performing the method |
| CN116596950A (en)* | 2023-05-31 | 2023-08-15 | 东北林业大学 | Retina fundus blood vessel tracking method based on feature weighted clustering |
| CN116596950B (en)* | 2023-05-31 | 2023-11-17 | 东北林业大学 | Retinal fundus blood vessel tracking method based on feature weighted clustering |
| CN116721760A (en)* | 2023-06-12 | 2023-09-08 | 东北林业大学 | Biomarker-fused multitasking diabetic retinopathy detection algorithm |
| CN116721760B (en)* | 2023-06-12 | 2024-04-26 | 东北林业大学 | A multi-task diabetic retinopathy detection algorithm integrating biomarkers |
| Publication number | Publication date |
|---|---|
| CN106127849B (en) | 2019-01-11 |
| Publication | Publication Date | Title |
|---|---|---|
| CN106127849A (en) | Three-dimensional fine vascular method for reconstructing and system thereof | |
| CN110782474B (en) | Deep learning-based method for predicting morphological change of liver tumor after ablation | |
| US6813373B1 (en) | Image segmentation of embedded shapes using constrained morphing | |
| CN103310458B (en) | In conjunction with the medical image elastic registrating method of convex closure coupling and multiple dimensioned hierarchical policy | |
| Niessen et al. | Multiscale segmentation of three-dimensional MR brain images | |
| CN113112609A (en) | Navigation method and system for lung biopsy bronchoscope | |
| CN109478327B (en) | A method for automatic detection of systemic arteries in any field of view computed tomography angiography (CTA) | |
| CN107292312B (en) | Tumor CT image processing method | |
| CN106780518A (en) | A kind of MR image three-dimensional interactive segmentation methods of the movable contour model cut based on random walk and figure | |
| CN108053433A (en) | A kind of multi-modal arteria carotis MRI method for registering based on physical alignment and outline | |
| Wu et al. | Segmentation and reconstruction of vascular structures for 3D real-time simulation | |
| CN116433734A (en) | Registration method for multi-mode image guided radiotherapy | |
| Wei et al. | Learning-based 3D surface optimization from medical image reconstruction | |
| CN102903117A (en) | 3D (three-dimensional) image registration method and device based on conformal geometric algebra | |
| CN119359930B (en) | Cerebral blood vessel three-dimensional model construction method and system based on cerebral blood vessel image segmentation | |
| CN107516314A (en) | Medical image supervoxel segmentation method and device | |
| Ebrahimdoost et al. | Automatic segmentation of pulmonary artery (PA) in 3D pulmonary CTA images | |
| CN106023094A (en) | Image-based bone tissue microstructure restoration system and restoration method thereof | |
| Zhou et al. | The line-and block-like structures extraction via ingenious snake | |
| Rahmawati et al. | Modification Rules for Improving Marching Cubes Algorithm to Represent 3D Point Cloud Curve Images. | |
| Kuba et al. | Medical applications of discrete tomography | |
| Horn et al. | Robust skeletonization for plant root structure reconstruction from MRI | |
| Bosc et al. | Statistical atlas-based sub-voxel segmentation of 3D brain MRI | |
| Commowick et al. | An efficient locally affine framework for the registration of anatomical structures | |
| Peng et al. | 3D segment and pickup framework for pancreas segmentation |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee | Granted publication date:20190111 Termination date:20210510 |