CN107689043B - Method for acquiring blood vessel section terminal node and branch node - Google Patents

Abstract

The invention belongs to the field of medical image processing technology, and in particular relates to a method for acquiring terminal nodes and branch nodes of a blood vessel segment. According to the connectivity relationship of these skeleton branch nodes, when the branch nodes have adjacent relationships, a branch aggregation is formed; the skeleton relationship is reconstructed through the branch nodes and connecting edges, and the above steps are repeated until the number of connected neighbors of the branch nodes is greater than 3 or the branch The number of nodes is no longer reduced; the present invention better builds the relationship between the skeleton branch nodes and other skeleton points, and provides a basis for measuring the length of the blood vessel segment and the radius of the blood vessel.

Description

A method for acquiring terminal node and branch node of blood vessel segment

Technology neighborhood

The invention belongs to the technical field of medical image processing, and particularly relates to a method for acquiring terminal nodes and branch nodes of a blood vessel segment.

Background technique

In recent years, with the rapid development and popularization of new imaging technologies and equipment such as computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET), the world's A large number of research images are generated every day in various places, which makes the use of research images for organ, tissue and blood vessel analysis one of the current research hotspots. The acquisition of blood vessels has important guiding significance for the analysis of individual blood vessels. Through the automatic detection of nodules in the image, the correctly segmented blood vessel structure can be used to analyze the ambiguity of the lung tissue structure in the same region.

In low-dose imaging, in addition to the more complex tissue or organ, the vascular image is affected by a large amount of noise from imaging and partial volume effects. This makes the contrast of blood vessels and other tissues low, so that it is difficult for traditional image blood vessel segmentation methods to obtain good segmentation results.

At present, many scholars at home and abroad have been devoted to the research of various medical image segmentation algorithms for many years to implement the segmentation of blood vessels in images, although these existing methods can obtain certain lung segmentation effects under certain conditions. However, due to the multi-level branch structure of blood vessels, the vascular tree structure segmented by the usual segmentation methods mostly lose many small blood vessel branches. In low-dose images, the contrast between blood vessels and other lung tissues is usually low; The influence of noise will cause the segmented blood vessels to have a large number of broken or lost vascular branches that should be connected. Therefore, it is difficult for such a segmentation method to obtain a complete vessel tree structure, which makes it lack quantification capability and cannot provide parameter information of specific vessel lengths and corresponding diameters. The fundamental reason is that there is a lack of basic means to process and analyze the segmented blood vessels, and most methods only provide qualitative image display, lack of accurate skeleton analysis methods for specific blood vessel cases, and due to the discrete nature of the blood vessel skeleton itself, it is difficult to accurately To determine the exact direction of the blood vessels, it is difficult to obtain better analysis and segmentation results.

SUMMARY OF THE INVENTION

Aiming at the deficiencies of the prior art, the present invention proposes a method for acquiring terminal nodes and branch nodes of a blood vessel segment, which aims to provide a basis for improving the effective method for analyzing and measuring blood vessels by images. The skeleton analysis based on graph analysis provides a basis for measuring the length and diameter of blood vessels. Because of its certain mathematical foundation and better realization means, this method has wide application prospects.

A method for obtaining a terminal node and a branch node of a blood vessel segment joint according to the present invention, as shown in FIG. 1 , includes:

S1. Analyze the neighborhood relationship of skeleton points to obtain a skeleton relationship set;

S2, reduce the pseudo-skeleton branch;

S3, analyzing the skeleton branch nodes S _b , and according to the connectivity relationship of these skeleton branch nodes, when the branch nodes have adjacent relationships, a branch aggregation is formed;

S4. Rebuild the skeleton relationship through branch nodes and connecting edges, and repeat S1, S2 and S3 until the number of connected neighborhoods of the branch node is greater than 3 or the number of branch nodes no longer decreases.

Preferably, step S0 is also included before step S1. As shown in FIG. 2, noise reduction is performed on the blood vessels obtained by segmentation according to the connectivity of the blood vessels, including:

Find connected voxels according to the neighborhood relationship, and split the blood vessel voxels into several parts;

Calculate the size of each connected body, and select the largest connected body to be the connected blood vessel;

The connected blood vessel is used as the blood vessel mask and the central skeleton voxel is multiplied to remove the skeleton noise;

Perform isotropic interpolation on blood vessels with skeleton noise removed, record the size of each voxel, and use mature skeletonization methods to complete the blood vessel skeleton.

Further, the isotropic interpolation includes:

The ratio of the maximum value to the middle value and the ratio of the maximum value to the minimum value of the coordinates of the skeleton point are used as the interpolation ratio, and the coordinates of the skeleton point obtained by the equal square interpolation on the x-axis, the y-axis, and the z-axis are the same. .

Preferably, the skeleton point neighborhood relationship is analyzed to obtain a skeleton relationship set including:

From the obtained skeleton points, obtain the skeleton point neighborhood points or skeleton point values;

Starting from any skeleton point, and cutting the ball with the neighborhood, searching for skeleton points in the neighborhood of this point, and counting them as the number of neighborhood points N _b ;

When the number of neighbor points N _b is greater than or equal to 3, the skeleton branch node is merged into the branch node set S _b ; when the number of neighbor points N _b is equal to 2, it is a common connected skeleton point set S _c ; when the number of neighbor points N _b is equal to 1, it is the terminal skeleton point set S _t .

Preferably, the calculating the number of neighborhoods of the data includes: for any point _p belonging to the branch node set Sb, the number of neighborhood points _Nb is the sum of the norms of all numbers in the _N26 neighborhood relationship.

Preferably, the reducing the pseudo-skeleton branch comprises:

According to the anatomical knowledge of the processed blood vessels, the radius _Ri of the blood vessels at all levels is set, and the wrong skeletonized branches are reduced from the candidate terminal branch skeleton B _t ;

Set the pruning skeleton pseudo-branch decision threshold L _r according to the number of branch skeleton points N(B _t ) and the radius characteristics; when N(B _t ) is less than L _r , it is regarded as a pseudo-branch, and the pseudo-branch is removed and the neighborhood structure of the skeleton point is updated .

Further, the branch aggregation includes:

Analyse the adjacent skeleton branch aggregation points, when more than two or more branch nodes are adjacent to each other, a branch node aggregation is formed, and each aggregation number N _c is allocated to find the aggregation center S _c ;

Any skeleton point belonging to S _b , if p belongs to the neighborhood of any other skeleton point, forms a cluster C, and its cluster center is Pc-(x _c , y _c , z _c ).

Preferably, it is characterized in that the calculation of the aggregation center Pc-(x _c , y _c , z _c ) includes:

Among them, n _c is the number of branch aggregations, and _xi , _yi , and _zi are the coordinates of each neighborhood skeleton point in aggregation C, respectively.

Preferably, the skeleton relationship is reconstructed through branch nodes and connecting edges, and S1, S2 and S3 are repeated until the number of connected neighborhoods of the branch nodes is greater than 3 or the number of branch nodes is no longer reduced, including:

Rebuild the skeleton relationship by branching nodes and connecting edges;

After removing too short pseudo branches and wrong branches, the neighborhood structure of branch nodes will be changed, so the skeleton relationship needs to be updated;

After updating the skeleton relationship, reconstruct the skeleton number according to the vessel segment set, and re-analyze the skeleton point neighborhood relationship, that is, repeat S1, S2 and S3 until the number of connected neighborhoods of branch nodes is greater than 3 or the number of branch nodes no longer decreases.

Compared with the prior art, the present invention provides a method for acquiring terminal nodes and branch nodes of a blood vessel segment. The blood vessel voxels segmented by an image are used to first perform noise removal, then perform isotropic interpolation, and then perform skeletonization. The neighborhood relationship between points and the geometric relationship between the central path of the blood vessel and the blood vessel, the terminal nodes and branch nodes are analyzed, and the relationship between the image space and the physical space of the divided blood vessels is used, so it is better to construct the skeleton and branch nodes. The relationship of other skeleton points.

Description of drawings

1 is a flowchart of a preferred embodiment of a method for obtaining an end node and a branch node according to the present invention;

2 is a flowchart of another preferred embodiment of the method for obtaining an end node and a branch node according to the present invention;

3 is a schematic diagram of the skeleton point and skeleton analysis of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and Not all examples.

A method for obtaining a terminal node and a branch node of a blood vessel segment proposed by the invention, as shown in Figure 1 and Figure 2, specifically includes the following steps:

S0, reduce the noise of the blood vessels obtained by segmentation according to the connectivity of the blood vessels

Find connected voxels according to the _N26 neighborhood relationship, and split the blood vessel voxels into several parts;

Calculate the size of each connected body, and select the largest connected body as the connected blood vessel;

The connected blood vessels are used as blood vessel masks to perform AND operation with the central skeleton voxel to remove the skeleton noise;

Perform isotropic interpolation on blood vessels with skeleton noise removed, record the size of each voxel, and use mature skeletonization methods to complete the blood vessel skeleton.

Further, the blood vessel mask refers to binarizing the obtained blood vessel image, that is, the part containing the connected blood vessel is marked as "1", and the part without the connected blood vessel is marked as "0".

Further, the method for removing the skeleton noise is to perform AND operation on the blood vessel mask and the central skeleton voxel, and remove the part that does not contain the blood vessel connected voxels, that is, remove the skeleton noise; specifically:

V _vessels = V _s &Mask _v

where V _vessels is the denoised vessel segment, Mask _v is the vessel mask, and V _s is the central skeleton voxel.

Further, the isotropic interpolation includes:

Denote the segmented and denoised blood vessel voxels as S(S _x , S _y , S _z ), S _x , S _y , S _z are the scales of blood vessels, where the maximum value of S _x , S _y , S _z is the same as the middle The ratio of the value and the ratio of the maximum value to the minimum value are used as the interpolation ratio, and the _{isotropic interpolation satisfies} the equal _dimension of the skeleton point. S _x , S _y , S _z ), S _med = med(S _x , S _y , S _z ), the interpolation ratio is: S _max /S _med , S _max /S _min , and the isotropic interpolation result satisfies: S' _x = S' _y = S' _z , that is, the iso-directional interpolation is to ensure the isotropy of the blood vessels after segmentation and denoising, and reduce false branches generated by skeletonization.

S1. Analyze the neighborhood relationship of skeleton points to obtain a set of skeleton relationships

Preferably, by analyzing the neighborhood relationship of skeleton points, as shown in FIG. 3 , the obtained skeleton relationship set includes:

From the obtained skeleton points, obtain the skeleton point neighborhood points or skeleton point values;

Starting from any skeleton point, and cutting the ball with the neighborhood, searching for skeleton points in the neighborhood of this point, and counting them as the number of neighborhood points N _b ;

When the number of neighbor points N _b is greater than or equal to 3, the skeleton branch node is merged into the branch node set S _b ; when the number of neighbor points N _b is equal to 2, it is the common skeleton connected point set S _c ; when the number of neighbor points N _b is equal to 1, it is the terminal skeleton point set S _t .

Among them, v _b is the skeleton point set.

The relationship between points, edges and neighborhoods that can be obtained through this step is as follows:

S2, reduce the pseudo-skeleton branch

Preferably, the reducing the pseudo-skeleton branch comprises:

According to the anatomical knowledge of the processed blood vessels, the radius _Ri of the blood vessels at all levels is set, and the wrong skeletonized branches are reduced from the candidate terminal branch skeleton B _t ;

Set the pruning skeleton pseudo-branch judgment threshold L _r according to the number of branch skeleton points N(B _t ) and the radius characteristic;

When N(B _t ) is smaller than L _r , it is considered a pseudo branch, and the pseudo branch is removed and the neighborhood structure of the skeleton point is updated.

Further, the pseudo branch determination threshold L _r is the ratio length tR of the blood vessel diameter of the current branch to the blood vessel voxel length, and L _r may be a ratio of 1.5-2 times the current rough estimated radius R.

For example, e belongs to the terminal edge set E _t , if L(e)≤tL _r , remove the terminal edge, for the branch node set S _b belonging to e, N(s _b )=N(s _b )-1, where N (s _b ) is the number of neighbor points of the branch node set S _b .

S3. Analyze the skeleton branch nodes S _b , and according to the connectivity relationship of these skeleton branch nodes, when the branch nodes have adjacent relationships, a branch aggregation is formed

Preferably, the branch aggregation includes:

The adjacent skeleton branch aggregation points are analyzed. When more than two branch nodes are adjacent to each other, a branch node aggregation is formed, and each aggregation number N _c is allocated to find the aggregation center;

Any skeleton point belonging to S _b , if p belongs to the neighborhood of any skeleton point, forms an aggregation C, and its aggregation center is Pc-(x _c , y _c , z _c ), using the aggregation center Pc-(x _{c )} ,y _c ,z _c ) to refine branch points and remove false branch points.

Preferably, it is characterized in that the calculation of the aggregation center Pc-(x _c , y _c , z _c ) includes:

Among them, n _c is the number of branch aggregations, and _xi , _yi , and _zi are the coordinates of each neighborhood skeleton point in aggregation C, respectively.

S4. Rebuild the skeleton relationship through branch nodes and connecting edges, repeat S1, S2 and S3 until the number of connected neighbors of the branch node is greater than 3 or the number of branch nodes is no longer reduced

Preferably, the calculation of the number of neighborhoods in the data includes: for any point _p belonging to the branch node set Sb, the number of neighborhood points _Nb is the sum of all numbers in the _N26 neighborhood relationship, that is,

Preferably, the skeleton relationship is reconstructed through branch nodes and connecting edges, and S1, S2 and S3 are repeated until the number of connected neighborhoods of the branch nodes is greater than 3 or the number of branch nodes is no longer reduced, including:

Rebuild the skeleton relationship by branching nodes and connecting edges;

After removing too short pseudo branches and wrong branches, the neighborhood structure of branch nodes will be changed, so the skeleton relationship needs to be updated;

After updating the skeleton relationship, reconstruct the skeleton number according to the vessel segment set, and re-analyze the skeleton point neighborhood relationship, that is, repeat S1, S2 and S3 until the number of connected neighborhoods of branch nodes is greater than 3 or the number of branch nodes no longer decreases.

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium can include: ROM, RAM, magnetic disk or optical disk, etc.

The above-mentioned embodiments further describe the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above-mentioned embodiments are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made to the present invention within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (8)

1. A method for acquiring a blood vessel section terminal node and a branch node is characterized by comprising the following steps:

s0: the noise reduction of the segmented blood vessels according to the connectivity of the blood vessels comprises the following steps:

searching connected voxels according to the neighborhood relationship, and dividing vessel voxels into a plurality of parts;

calculating the size of each communicating body, and selecting the largest communicating body as a communicated blood vessel;

taking the communicated blood vessel as a blood vessel mask to be multiplied with the central skeleton voxel, thereby removing skeleton noise;

performing isotropic interpolation on the blood vessel without skeleton noise, recording the size of each voxel, and finishing the blood vessel skeleton by using a skeletonization method;

s1, analyzing the neighborhood relationship of the skeleton points to obtain a skeleton relationship set;

s2, cutting pseudo skeleton branches;

s3, analyzing framework branch node S_bAccording to the connection relation of the framework branch nodes, when the branch nodes have adjacent relation, branch aggregation is formed;

s4, the skeleton relationship is rebuilt through the branch nodes and the connecting edges, and the S1, the S2 and the S3 are repeated until the number of the connection neighborhoods of the branch nodes is larger than 3 or the number of the branch nodes is not reduced any more.

2. The method according to claim 1, wherein the isotropic interpolation comprises:

and taking the ratio of the maximum value to the intermediate value and the ratio of the maximum value to the minimum value of the coordinates of the skeleton points as interpolation ratios, wherein the coordinates of the skeleton points obtained by the equi-square interpolation on the x axis, the y axis and the z axis are the same.

3. The method according to claim 1, wherein the analyzing the neighborhood relationship of the skeleton points to obtain the skeleton relationship set comprises:

obtaining framework point neighborhood points from the obtained framework points;

starting from any skeleton point, using neighborhood cutting ball to search skeleton point in neighborhood of the point, and counting the skeleton point as neighborhood point number N_b；

Number of current neighborhood points N_bIf the number of the branch nodes is more than or equal to 3, the branch nodes of the framework are merged into a branch node set S_b(ii) a Number of current neighborhood points N_bEqual to 2, then is a common connected skeleton point set S_c(ii) a Number of current neighborhood points N_bEqual to 1, the terminal skeleton point set S is_t。

4. The method according to claim 3, wherein the calculating of the number of neighborhood points comprises: for any belonging to the branch node set S_bP, number of neighborhood points N_bIs N₂₆The sum of the norms of all numbers in the neighborhood.

5. The method according to claim 1, wherein the reducing of the pseudo skeleton branches comprises:

setting the radius R of each level of blood vessel according to the anatomical knowledge of the treated blood vessel_iBranching skeleton B from candidate terminal_tReducing erroneous skeletonized branches;

according to branch skeleton number N (B)_t) And setting a trimming framework pseudo branch judgment threshold L according to the radius characteristic_r；

When N (B)_t) Less than L_rAnd if the frame is a pseudo branch, removing the pseudo branch and updating the neighborhood structure of the skeleton point.

6. The method according to claim 1, wherein the branch aggregation comprises:

analyzing the adjacency of the skeleton branch aggregation points, forming branch node aggregation when more than two branch nodes are adjacent, and distributing each aggregation number N_cFinding the center of aggregation S_c；

Belonging to a set of branch nodes S_bThe neighborhood skeleton points of any skeleton point of (a) form an aggregate C with an aggregate center of Pc.

7. The method of claim 6, wherein the calculating the coordinates of the c + Pc center comprises:

wherein x is_c,y_c,z_cX-axis coordinate, Y-axis coordinate, and Z-axis coordinate, n, respectively, of the focus center Pc_cIs the number of branch aggregations, x_i、y_i、z_iRespectively, coordinates of each neighborhood skeleton point in the aggregation C.

8. The method of claim 1, wherein the reconstructing the skeleton relationship by the branch nodes and the connecting edges and repeating the steps S1, S2 and S3 until the number of connection neighborhoods of the branch nodes is greater than 3 or the number of branch nodes is not reduced further comprises:

reconstructing a skeleton relationship through the branch nodes and the connecting edges;

updating the skeleton relationship;

and reconstructing the skeleton number according to the blood vessel segment set, and re-analyzing the skeleton point neighborhood relationship, namely repeating S1, S2 and S3 until the connection neighborhood number of the branch nodes is more than 3 or the branch node number is not reduced any more.

Images