KR100315422B1 - Method of abstraction and segmentation of 3D features of blood vessels from medical volume data for surgical simulation - Google Patents

Abstract

The present invention relates to a method of extracting and abstracting geometric three-dimensional information about the blood vessels of a human body used in medical surgery simulation from medical stereoscopic data, and to a computer-readable recording medium recording a program for realizing the method. We introduce the concept of planar wave, extract three-dimensional information on the blood vessels of the human body from medical image volume data, approximate them into generalized cylinders, abstract them, and store them in a symbolic data structure. In order to provide a recording computer-readable recording medium, the first step of extracting the geometric three-dimensional information about the blood vessels of the human body required in the tube insertion radiation surgery simulation from the medical image volume data using a plane wave; A second step of approximating and abstracting geometric three-dimensional information about the blood vessels of the human body into a cylinder using a generalized plane wave; And a third step of storing the approximated and abstracted geometric three-dimensional information in a symbolic data structure, and used for a medical surgery simulation system.

Description

Method of abstraction and segmentation of 3D features of blood vessels from medical volume data for surgical simulation}

The present invention relates to a method of extracting and abstracting geometric three-dimensional information about the blood vessels of a human body used in medical surgery simulation from medical stereoscopic data, and to a computer-readable recording medium recording a program for realizing the method. In particular, the extraction and abstraction method for extracting the three-dimensional geometric information of the blood vessels required by the tube implant surgery simulator using the virtual reality (VR) technology from the medical image volume data, and the program for realizing the method are recorded. The present invention relates to a computer-readable recording medium.

1 is an exemplary configuration diagram of a general computer.

1 shows a general computer, which includes an input unit 11, a central processing unit 12, an output unit 13, a storage unit 14, and the like. The present invention relates to a method for operating in a medical surgery simulation system constructed based on such a general computer.

As virtual reality technology has been applied to the medical field, many medical simulation systems have been developed and are currently used as an aid to medical practice in many fields. In particular, in the case of surgery, the application is becoming active, mainly used for intern training or mock surgery before the actual surgery.

In the present invention, such a surgical simulator is targeted to a Vascular and Interventional Radiology Simulation System (VIRSS) that simulates the process of treating heart attack. This system simulates the entire process of surgery such as inserting a tube into the blood vessel of the human body to prevent clogging around the heart, or inserting a filter or injecting a laser when the blood vessel is blocked to penetrate the blockage. It is mainly used for intern training. This is because such a surgical procedure is dangerous if the actual patient is targeted for internship training, and it is difficult to obtain the patient.

The above simulation system requires three important functions or operations.

First, real-time visualization is needed to show the preoperative procedure, such as the tube being inserted through the vein.

Second, dynamics simulation is needed to simulate the insertion of the tube into the vein and give the user feedback.

Finally, a collision detection between the tube and the blood vessel wall is needed to examine the tube being inserted into the blood vessel wall.

However, it is difficult to efficiently perform all three operations with one data type, such as a polygonal surface for blood vessels.

A symbolic model should be created using the extracted geometrical information, and then converted into an appropriate data structure as needed to perform the required operation.

Therefore, abstract the geometric 3D information on the blood lines to form a tree shape and convert it into a polygonal surface or an implicit surface as needed to efficiently perform the above functions. .

Extracting three-dimensional geometric information about blood vessels from medical images is a difficult task. Blood Vessels are almost as dense as other organs in the human body, such as muscles, lungs, and liver, and because most of the currently developed medical devices measure the density of objects, it is difficult to distinguish between blood and other organs in the resulting image. Because. To overcome this, a recently developed technique called angiography has been developed, which acquires an image after scanning a dye, which reacts well to radiation, onto a blood vessel before the image is acquired. This makes the information about your veins stronger than other organs. It is assumed that the data used as an input in the present invention mainly uses medical image volume data obtained through the angiography method.

The most representative algorithm for extracting three-dimensional information about blood vessels from medical image volume data is the Marching Cubes algorithm. This is very efficient for extracting surface information about blood vessels. However, the extracted surface is a set of triangles and retains only the three-dimensional information of the triangle, so that geometric and hierarchical information such as branches, radii, and medial axis of the veins can be obtained. It is efficient to visualize the simulation process because it does not have it, but it is not efficient for dynamic simulation or collision inspection between blood lines and pipes. In order to overcome this, a wave propagation method has been developed.

2 is a flowchart illustrating a conventional wave propagation method.

In the conventional wave propagation method, when the user gives an initial position in the blood line (201), sets the initial wave using the seed as the seed (202), generates the wave repeatedly, and then propagates the wave. It finds all the blood lines associated with the initial position (203, 204). Here, a wave is a collection of voxels that are connected to a given seed point (usually 26-connected) and constitute blood vessels, as the ripple spreads around the point where the stone fell when it was thrown on the surface. It can be expressed as And, wave propagation refers to generating the next wave based on the current wave, which does not belong to the current wave among the voxels connected to each voxel of the current wave, A set of corresponding things.

This wave propagation method finds all the blood lines associated with the seed point, which is especially useful when the blood lines have branches or bifurcation. That is, since branch detection occurs when a wave has two separate components during wave propagation, branching points can be easily detected through the connected component analysis used in computer vision. have.

However, in spite of these advantages, there is a fatal disadvantage of this method, because the propagating dive is not parallel to the cross-section of the vein, which requires a large number of images for its processing. This is because when a wave is spreading and propagating, it may cause erratic results.

The present invention was devised to solve the above problems, and introduced the concept of plane wave to extract three-dimensional information on the blood vessels of the human body from the medical image volume data, approximated to a generalized cylinder, and then abstracted the symbolic form. It is an object of the present invention to provide a computer readable recording medium storing a method for storing the data structure and a program for realizing the method.

1 is an exemplary configuration diagram of a general computer.

Figure 2 is a flow diagram of a conventional wave propagation method.

3A to 3C are conceptual views illustrating a method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body used in medical surgery simulation according to the present invention from stereoscopic data of medical images.

4 is a flow diagram of an embodiment of a process of extracting a blood vessel from volume data and simultaneously approximating it by a generalized cylinder in accordance with the present invention.

5A and 5B illustrate an embodiment of a process of extracting a blood vessel from volume data and approximating the same by a generalized cylinder according to the present invention.

6a and 6b is an embodiment conceptual diagram for the process of detecting the branch in the extraction of the blood vessels of the human body according to the present invention.

FIG. 7 is a flowchart illustrating a method of extracting and abstracting geometric three-dimensional information about a blood vessel of a human body used in medical surgery simulation according to the present invention from three-dimensional data of a medical image. FIG.

* Explanation of symbols for the main parts of the drawings

11 input unit 12 central processing unit

13 output unit 14 storage unit

The present invention for achieving the above object, in the method for extracting and abstracting the geometric three-dimensional information on the blood vessels of the human body to be applied to the medical surgery simulation system in the medical image stereoscopic data, which is required in the tube implantation radiation surgery simulation A first step of extracting geometric three-dimensional information on the blood vessels of the human body from the medical image volume data using a plane wave; A second step of approximating and abstracting geometric three-dimensional information about the blood vessels of the human body into a cylinder using a generalized plane wave; And a third step of storing the approximated and abstracted geometric three-dimensional information in a symbolic data structure.

In addition, the present invention is a medical surgery simulation system having a large-capacity processor, the first to extract geometric three-dimensional information on the blood vessels of the human body required in the tube implantation radiation surgery simulation from the medical image volume data using a plane wave function; A second function of approximating and abstracting geometric three-dimensional information about the blood vessels of the human body into a cylinder by using a generalized plane wave; And a computer readable recording medium storing a program for realizing a third function of storing approximated and abstracted geometric three-dimensional information in a symbolic data structure.

The present invention has introduced a new concept called a plane wave to improve the advantages and improve the disadvantages of the wave propagation method described above, and to simply extract the three-dimensional information about the blood vessels of the human body from the medical image volume data Instead, it is approximated to a generalized cylinder, abstracted, and stored as a symbolic (tree structure) data structure, and if necessary, the structure is converted to another structure to perform an efficient operation.

Plane wave is a collection of voxels of waves formed by limiting wave propagation to a given voxel on a given plane. When the plane equation is given perpendicular to the axial direction of the blood line, the plane wave is formed by cross-section of the blood line. Parallel. This property is later useful for calculating geometric information about the blood vessels (radius, central axis).

The approximation of the blood line by the generalized cylinder is based on the curvature of the blood line. In the large curvature, the interval of the cylinder is narrow, and in the small curvature, the interval of the cylinder is widened to approximate.

In the present invention, a method of estimating the curvature of the blood lines shoots a ray from the center of the plane wave in the axial direction of the blood lines, where the distance traveled by the ray is inversely related to the curvature of the blood lines. This relationship is used to determine where to create a new node (one end of the cylinder). The axial direction of the vein is easily determined by propagating the plane wave once and then finding the vector connecting the centers of the two plane waves.

Each node in the final tree has information such as the three-dimensional coordinates of the point, the radius of the veins at that point, the presence or absence of branches, and the medial axis.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3A to 3C are conceptual views illustrating a method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body used in medical surgery simulation according to the present invention from three-dimensional data of a medical image.

3A to 3C illustrate a process of extracting blood vessels from medical image volume data, approximating them by generalized cylinders, and storing them as abstracted tree data structures.

FIG. 3A is an example of medical image volume data, FIG. 3B is an example of approximation by a generalized cylinder of a blood vessel of the human body, and FIG. 3C is an example of an abstracted tree data structure. It is also.

3A to 3C, the tree data structure abstracted from the medical image volume data is generated.

The present invention proposes a method for extracting three-dimensional geometric information of a blood vessel required by a tube insertion surgery simulator using virtual reality (VR) technology from medical image volume data. At this time, the extracted blood line is approximated by a generalized cylinder based on the curvature, and the generated cylinders are converted into a tree data structure while maintaining the original hierarchical structure. The overall composition of this process can be divided into a module for extracting blood vessels from volume data and at the same time a module for approximating by generalized cylinders and a branch for blood vessels.

4 is a flow chart of an embodiment of the process of extracting the blood vessels from the volume data according to the present invention and simultaneously approximating them by a generalized cylinder.

First, referring to the approximation process with reference to FIG. 4, if it is assumed that a new node is created first, the position of the new node is determined (401), and geometric information of the blood lines corresponding to the node is obtained and given as the attribute. (402). At this time, information such as the radius of the blood line, the axial direction, the branch, etc. are stored in the node.

Next, after finding the axial direction of the vein at the center of the planar wave (403), the ray is fired to measure the curvature of the vein based on the distance the ray has traveled in the vein (404).

It is checked whether the distance traveled by Ray is less than or equal to epsilon (405).

If the distance traveled by the ray is less than or equal to epsilon, the algorithm terminates because it indicates that there are no more blood lines, otherwise the process repeats from step 401 where the location of the new node is determined. Is approximated by a generalized cylinder and finally a tree-shaped data structure is obtained.

5A and 5B are conceptual views illustrating a process of extracting a blood vessel from volume data and approximating by a generalized cylinder at the same time according to the present invention.

5A and 5B are conceptual views showing the algorithm of FIG. 4. In FIG. 5A, the distance in which the ray moves is determined by generating a plane wave and determining the axial direction of the vein due to its propagation, and shooting a ray from the center of the plane wave. It shows the process of creating a new node repeatedly.

5B shows the resulting generalized cylinder.

6A and 6B are diagrams illustrating an example of a process of detecting a branch from an extraction of blood vessels of the human body according to the present invention.

Secondly, referring to the branch detection algorithm with reference to the branch detection diagrams of FIGS. 6A and 6B, when the algorithm of FIG. 4 is repeatedly performed, the branch detection algorithm may be ignored even if there is a branch of the blood vessel. To compensate for this, a branch detection algorithm was additionally developed, which determines the new node by the algorithm of FIG. 4, divides the old node with the new node at regular intervals (FIG. 6a), and then generates a planar wave to radii at each point. Will be investigated.

Then, calculate the difference between the radiuses of the adjacent points, assuming that there is a branch between them if the difference is very large, test the difference between the radiuses of the two sections by repeatedly bisectioning the two sections. The same test is repeated by setting a section having a large difference as a new section (FIG. 6B). In this process, the interval between sections where branches can exist is gradually narrowed, and finally, if the interval between the sections is below a certain value, the connected component analysis is performed while stopping the repetitive bisection and propagating the planar wave. In the analysis, the branch is detected by continuously examining the number of independent components.

FIG. 7 is a flowchart illustrating a method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body used in medical surgery simulation according to the present invention from three-dimensional image data.

7 is a very simple description of the branch detection part, which combines the two processes presented through the above description, and introduces a data structure such as a stack so that the algorithm operates on the entire blood line of the input data. . In this case, when the approximation algorithm encounters a branch during execution, two separate component wave forms exist, one of which continues to perform the approximation algorithm and the other is put on the stack to start processing the branch later. Will be written as

Referring to FIG. 7, a method of extracting a blood line from the medical image volume data proposed by the present invention and approximating it by a generalized cylinder and generating an abstracted data structure using structural and hierarchical information between cylinders is as follows. .

1. The user initially specifies a point in the vein as a Seed Position or Seed Voxel.

2. Propagate the wave from the seed point to form an initial plane wave, and propagate the formed plane wave only once to produce the next plane wave.

3. Find and connect the centers of the two planar waves previously obtained to find the direction of the axis of the vein. In this case, the center of the plane wave may be obtained by averaging three-dimensional coordinates of the voxels of the plane wave.

4. Shoot the ray from the center of the plane wave using the obtained axial direction. Ray travels in the veins and stops when he finally meets the veins wall.

5. The distance traveled by Ray is inversely proportional to the curvature of the vein, thus determining the position of the new node.

6. Once the new node has been located, create a planar wave at that location and obtain information about the radius of the vein. Then go to section 2 to loop.

7. If Ray traveled very little in Section 5, stop because there are no longer any lines along this branch, take another branch off the stack and repeat steps 2-5.

8. The process of detecting branches is inserted between the previously created node and the new node determined in clause 5. When a branch is detected, a new node is formed, and the planar wave is made of as many independent components as the number of branches, and they are all stored on the stack.

9. Then we take a plane wave from the stack and continue extracting and approximating the branches of its associated veins.

10. The final termination of the algorithm is when the stack is empty and propagation of the planar wave in section 2 no longer occurs or the distance traveled by the ray in section 5 is very small.

This will be described according to the flow of the drawings.

First, the initial position of the seed in the vein is set (701). An initial plane wave is formed and the formed plane wave propagates to form the next plane wave (702).

It is confirmed that propagation of the plane wave is no longer made (703). As a result of the propagation of the planar wave, it is checked whether there is a new branch (704). As a result of the check, if there is a new branch, the plane wave is split (705) and new plane waves are inserted into the stack (706).

As a result of checking whether there is a new branch, if no new branch exists, a new plane wave is inserted into the stack (706).

Next, it is checked if the stack is empty (707). As a result of the inspection, if the stack is not empty, information about one plane wave is retrieved from the stack (708), the axial direction is estimated, and the ray is fired (709). Then, Ray measures the distance traveled in the vein (710) and generates a new node based on the curvature of the vein (711). Next, the formed plane wave is propagated to repeat the process of making the next plane wave (702).

As a result of checking whether the stack is empty, if it is empty, it extracts blood lines from medical image volume data using plane wave and ray shooting and approximates them by generalized cylinder to abstract the geometric information of blood lines. End the process of creating.

As a result of confirming that propagation of the plane wave is no longer performed, if it is not made anymore, the process is repeatedly performed from step 707 of checking whether the stack is empty.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and various substitutions, modifications, and changes can be made in the art without departing from the technical spirit of the present invention. It will be apparent to those of ordinary knowledge.

As described above, the present invention is not limited to simply extracting three-dimensional geometrical information about the blood vessels of the human body required by the vascular simulation system, and approximating and abstracting it with a generalized cylinder, and then symbolic ( Tree structure) can be stored as a data structure, and if necessary, the structure can be converted to another structure to perform an efficient operation, and the tree of blood lines obtained as a result of the present invention has a polygonal surface blood line to visualize the simulation process in real time. It is possible to construct a function and to express the function in an implicit surface for dynamic simulation or collision detection.

Claims (7)

In the method of extracting and abstracting geometric three-dimensional information about the blood vessels of the human body applied to the medical surgery simulation system from the medical image stereoscopic data, A first step of extracting geometric three-dimensional information on the blood vessels of the human body required for the vascular surgery simulation from the medical image volume data using a plane wave; A second step of approximating and abstracting geometric three-dimensional information about the blood vessels of the human body into a cylinder using a generalized plane wave; And Third step of storing approximated and abstracted geometric three-dimensional information in symbolic data structure A method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body applied to a medical surgery simulation system including a medical image in stereoscopic data. The method of claim 1, The symbolic form is A method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body, which is a tree structure, from three-dimensional data of a medical image. The method of claim 1, The plane wave, A method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body, characterized in that the wave is a collection of voxels of waves formed by limiting wave propagation to voxels on a given plane. The method according to any one of claims 1 to 3, The second step, A fourth step of setting the position of the seed value in the vein; Generating a planar wave and propagating it to generate a next planar wave and connecting the centers of the two planar waves to obtain the axial direction of the blood vessels; A sixth step of shooting a ray from the center of the obtained planar wave in the axial direction of the vein and estimating the curvature of the vein using the distance traveled by the ray in the vein and creating a new node based thereon; A seventh step of dividing the two nodes at regular intervals and generating a planar wave to find and compare the radius at each point to detect the branch of the blood vessel; And Eighth step to approximate and abstract cylinder information about human blood lines Extracting and abstracting geometric three-dimensional information about the blood vessels of the human body from the medical image stereoscopic data. The method of claim 4, wherein The seventh step, A method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body from the stereoscopic data of a human body, characterized in that dichotomy is used to narrow the interval of the interval when the section of the blood vessel may be determined. The method of claim 4, wherein The seventh step, Two branches exist when one of the extracts of the blood lines is encountered. One of them is placed on the stack and the processing continues with the other. When it is finished, it returns to the branches and extracts the information of the other branches in the stack and processes the remaining branches. A method of extracting and abstracting geometric three-dimensional information about blood vessels of a human body from medical image stereoscopic data. In a medical surgery simulation system with a large processor, A first function of extracting geometric three-dimensional information on the blood vessels of the human body required by the tube insertion radiation surgery simulation from the medical image volume data using a plane wave; A second function of approximating and abstracting geometric three-dimensional information about the blood vessels of the human body into a cylinder by using a generalized plane wave; And A third function of storing approximated and abstracted geometric three-dimensional information in a symbolic data structure A computer-readable recording medium having recorded thereon a program for realizing this.