KR101294858B1 - Method for liver segment division using vascular structure information of portal vein and apparatus thereof - Google Patents

Abstract

PURPOSE: A method for classifying liver segment using vascular structure information of portal vein and an apparatus thereof are provided to implement more objective and accurate reading, by automating the reading method. CONSTITUTION: Three-dimensional chest volume data is acquired (S210). Route tree is generated by using a Euclidian distance map (S220). End point, branching point and skeleton of liver vein are extracted using level set (S230). The position of the extracted branching point is corrected (S240). Liver segment is classified by using the branching pattern structure (S260). [Reference numerals] (AA) Start; (BB) Finish; (S210) Acquire three-dimensional chest volume data; (S220) Generate a route tree by using an Euclidian distance map; (S230) Extract an end point, a branching point and a skeleton of a liver vein; (S240) Correct the position of the extracted branching point; (S250) Analyze branching patterns in respect to the liver vein and extract a main blood vessel; (S260) Classify a liver segment by using a branching pattern structure of the main blood vessel

Description

Method for liver segment division using vascular structure information of portal vein and apparatus etc.

The present invention relates to a method and apparatus for classifying liver segments using vascular structure information of the liver context, and more particularly, to analyze the vascular structure of the CT ported liver context and classify segments to maximize the success rate of liver transplantation. The present invention relates to a liver segment classification method and apparatus using vascular structure information of the liver context.

Currently, about 10,000 people die from liver disease in Korea. However, liver disease is the best way to get a healthy human liver because there is no curable medicine. Since the first successful bio-transplantation transplantation operation at Seoul Asan Hospital in 1994, the recent trend is rapidly shifting from brain-lion transplantation to biotransplantation. Liver has 30% of its total volume, which makes it difficult to maintain its life, and its regeneration is so strong that it can be revived up to 80% within a month. Liver partial liver transplantation is the only method of essential treatment for end-stage cirrhosis, progressive chronic hepatitis, early liver cancer, congenital biliary obstruction, and Wilson's disease.

The demand for accuracy of donor suitability assessments is increasing rapidly with the increase in biotransplantation. At this time, the most important thing is to accurately measure the volume of the donor's liver segment, which is absolutely necessary in terms of the recipient's postoperative course and the donor's safety. The most important thing to ensure the donor's postoperative safety is the remaining liver volume after donation.If the donor area is too large and the liver tissue remaining in the donor does not meet the donor's required standard liver weight, hepatic dysfunction is increased after surgery. The problem is that it usually leaves more than 35 to 40% of the total liver volume. Therefore, the remaining liver volume must be between 35 and 40% in order to be judged as a suitable donor for biotransplant liver transplantation.

Preoperative information on the volume of the liver segment for determining donor suitability described above can be obtained from abdominal computed tomography (CT) images. Previously, volume measurements of the donor's liver segment have been performed manually by a physician. . For this purpose, the liver segment was manually divided, the area of the ablation surface for manually distinguishing the liver for each slice was multiplied by the slice thickness, and the sum was obtained to measure the volume of the liver segment.

Therefore, the process of manually dividing the liver segment takes a very long time, and the liver segment has to divide the blood supply from the major liver portal blood vessels in three dimensions, but the doctor's manual work is performed on the two-dimensional plane. Because of the limitations, there is a problem inherent in accuracy.

The background technology of the present invention is described in Republic of Korea Patent Publication No. 10-1126447 (2012. 03. 29 registration).

Accordingly, a technical problem of the present invention is to provide a method and apparatus for classifying liver segments using vascular structure information of the liver context which can maximize the success rate of liver transplantation by analyzing the vascular structure of the CT-photographed liver context and classifying segments. It is.

Liver segment classification method using the vascular structure information of the liver context according to the present invention, the step of obtaining the three-dimensional chest volume data through CT imaging, using the three-dimensional chest volume data to the points on the blood vessel boundary of the liver context Generating a path tree using a Eucladian distance map for the step, extracting the end point, branch point and skeleton of the liver context using a level set, correcting the position of the extracted branch point, the liver context Analyzing the branching pattern for and extracting the main blood vessels from the skeleton, and using the branching pattern structure of the main blood vessels comprising the step of separating the liver segment.

The generating of the path tree may include generating the eucladian distance map for all points at the blood vessel boundary of the liver context, and selectively connecting the inner voxels farthest from the blood vessel boundary to connect the path tree. It may include the step of generating.

The step of extracting the end point, branch point and skeleton of the liver context may be defined as a set of points having the same integer partial value of the time it takes for the level set to propagate, and the set of points having one adjacent set as the liver. Determining as an end point of the context, extracting skeletal information of the hepatic context by connecting back traced from the end point of the extracted hepatic context to a starting point, and determining a point where the extracted plurality of skeletons intersect as a branch point It may include the step.

Compensating the position of the extracted branch point, the step of calculating the sphere of the maximum size inscribed in the blood vessel wall of the liver context by expanding the virtual sphere around the adjacent candidate points around the extracted branch point, And correcting the position of the branch point to a point corresponding to the center point of the sphere of the largest size.

As shown in the following equation, the coordinates of the virtual sphere are added by adding the relative coordinates OT _{x (i)} , OT _{y (i)} , and OT _{z (i)} stored in the offset table to the center coordinates (P _candidate ) for the candidate points. (P _sphere ) can be calculated.

Here, P _candidate represents a coordinate (x ₀ , y ₀ , z ₀ ) corresponding to the center of the virtual sphere.

Analyzing the branching pattern for the liver context and extracting the main blood vessel structure, setting a branch point located at the inlet portion of the liver context as a parent node, the branch node connected to the parent node and the parent node Setting a level for the liver context; analyzing a branching pattern for the hepatic context in a tree form from the level information and skeleton information; and a eucladian distance map for the points at the boundary of the hepatic context. Predicting the diameter, and using the diameter and the threshold value of the hepatic portal vein can be distinguished from the main vessel and the secondary vessel, and removing the secondary vessel.

According to another embodiment of the present invention, a liver segmentation device may further include a volume data acquisition unit for acquiring three-dimensional chest volume data through CT imaging, and for the points on the blood vessel boundary of the hepatic portal vein using the three-dimensional chest volume data. Path tree generation unit for generating a path tree using a Eucladian distance map, a bone extraction unit for extracting end points, branches and skeletons of the liver context using a level set, and a part for correcting the position of the extracted branch points A branch position correcting unit, a branch pattern analyzing unit analyzing a branching pattern for the liver context and extracting a main blood vessel from the skeleton, and a liver segment separating unit classifying a liver segment using a branching pattern structure of the main blood vessel .

Thus, according to the present invention, since the liver volume of the specific segment of the liver can be accurately measured from the liver context image of the donor, it can be provided to an actual clinician to determine the suitability of the donor based on this.

Furthermore, medical forms have been altered by automating and verifying qualitative reading methods that rely on the manual hand of the current physician, resulting in more objective and sophisticated readings. In addition, by connecting these devices to existing three-dimensional CT devices or picture archive and communication systems, not only can they bring significant sales and exports of software, but also the competitiveness of domestic medical systems in the global market. Can be strengthened.

1 is a block diagram of a device for classifying liver segments using vascular structure information of the liver context according to an embodiment of the present invention.
2 is a flowchart illustrating a method for classifying liver segments using vascular structure information of the liver context according to an embodiment of the present invention.
3 is a view for explaining a process of forming a path tree according to an embodiment of the present invention.
4 illustrates a priority table used in a path tree generation process according to an embodiment of the present invention.
FIG. 5 is a diagram for describing a process of extracting a level set based vessel control point according to an exemplary embodiment of the present invention.
6 is an exemplary view showing a result of extracting a skeleton, a branch point, and an endpoint of a liver context according to an embodiment of the present invention.
7 is a view for explaining a branch point position correction process according to an embodiment of the present invention.
8 is a diagram illustrating two-dimensional coordinate classification according to a distance D in a first quadrant according to an embodiment of the present invention.
9 is a view for explaining the analyzed level of each sub-branch according to an embodiment of the present invention.
Figure 10 shows the anatomical structure of the liver left segment according to an embodiment of the present invention.
Figure 11 shows the anatomy of the right liver segment in accordance with an embodiment of the present invention.
12 is a view for explaining the two-dimensional projection positional relationship of liver right segments according to an embodiment of the present invention.
FIG. 13 is an exemplary diagram illustrating a distance-based liver segment classification modeling method according to a blood vessel structure of a liver context according to an embodiment of the present invention.
14 is an exemplary diagram showing a segment classification result of a left liver segment to which the present invention is applied.
15 is an exemplary diagram showing a segment classification result of a liver right segment to which the present invention is applied.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention.

1 is a block diagram of a device for classifying liver segments using vascular structure information of the liver context according to an embodiment of the present invention. As shown in FIG. 1, the liver segmentation apparatus 100 includes a volume data acquisition unit 110, a path tree generation unit 120, a skeleton extractor 130, a branch point position correction unit 140, and a branch pattern analysis. The unit 150 and the liver segment separator 160 are included.

First, the volume data acquisition unit 110 acquires 3D chest volume data by a CT technique. The path tree generation unit 120 generates a path tree using a Eucladian distance map of points at the blood vessel boundary of the liver context using 3D chest volume data, and uses the path tree to generate a centerline of the liver portal blood vessel. Extract

The skeleton extractor 130 extracts the end point, the branch point and the skeleton of the liver context using the level set, and the branch point position corrector 140 corrects the position of the extracted branch point.

The branch pattern analysis unit 150 analyzes the branching pattern for the liver context and extracts the main blood vessel, and the liver segment separator 160 classifies the liver segment using the branch pattern structure of the main blood vessel.

Hereinafter, an automatic liver segmentation method using blood vessel structure information of the liver context in computed tomography according to an exemplary embodiment of the present invention will be described with reference to FIGS. 2 to 13.

2 is a flowchart illustrating a method for classifying liver segments using vascular structure information of the liver context according to an embodiment of the present invention.

First, the volume data acquisition unit 110 acquires 3D chest volume data by CT imaging (S210). The three-dimensional chest volume data is three-dimensional volume data including the liver, and a method of acquiring image data using a CT technique is well known and a detailed description thereof will be omitted.

Next, the path tree generation unit 120 generates a path tree by using the Eucladian distance map of the points on the blood vessel boundary of the liver context (S220).

3 is a view for explaining a process of forming a path tree according to an embodiment of the present invention, Figure 4 shows a priority table used in the process of generating a path tree according to an embodiment of the present invention.

As shown in FIG. 3, the path tree generator 120 generates a path tree to generate a center tree connecting internal voxels having a local maximum having a local maximum value from the hepatic portal blood vessel boundary. The process is carried out in the following way. In other words, assuming that hepatic portal blood vessels are divided into parts in the longitudinal direction, a centerline is generated by finding and connecting voxels corresponding to the furthest points from the blood vessel boundary of each part, and a path tree is generated by expanding the centerline. .

In a first step, the path tree generator 120 generates a 3D Euclidean distance map for all the points on the blood vessel boundary of the liver context. In this case, the actual size of the voxel of the CT data in the x, y, and z-axis directions is taken into account when generating the distance map. First, as illustrated in FIG. 3A, the branch paths are sequentially predicted in advance, wherein the voxel distances in the x and y axis directions correspond to pixel spacing, and the voxel distances in the z axis direction correspond to slice intervals. Therefore, when generating the distance map, the path tree generator 120 multiplies the slice interval by the pixel spacing in the z-axis direction to correct the difference in the actual distance along the axial direction.

In a second step, the path tree generator 120 first visits the inner voxels farthest from the blood vessel boundary of the liver context to generate a path tree. The reason for preferentially visiting the inner voxels farthest from the vascular border of the hepatic portal vein is to ensure that the centerline includes a point that crosses the center of the hepatic portal portal as far as possible.

Here, the distance away from the blood vessel wall is defined as the weight of the node so that a node having a large weight has priority to propagate.

The path tree generator 120 accelerates the path tree generation process using the priority table of FIG. 4. The index of the priority table corresponds to the weight of the node defined by the distance from the vessel wall. As shown in FIG. 4, the nodes connected to each index not only have weights but also have a cumulative distance from the starting point, a pointer to the parent node, and node x, y, z coordinate information.

When a new voxel is visited during the path tree generation, the voxels are sorted according to weights and inserted into corresponding indexes. 3 (b) and 3 (c) show an example of an intermediate process in which neighboring points not previously visited from the starting point S are inserted into the priority table. That is, in FIG. 3B, node 3 is inserted, and in FIG. 3C, node 5 is inserted.

Here, the starting point is determined to be the point at which the distance from the vessel wall of the inlet is farthest, and the path tree is expanded by connecting the nodes with the maximum weights.

Next, the skeleton extracting unit 130 extracts the end point, branch point and skeleton of the liver context using the level set (S230).

FIG. 5 is a diagram for describing a process of extracting a level set based vessel control point according to an exemplary embodiment of the present invention.

First propagate a set of levels with velocity F (x) as shown in Equation 1 according to the Euclidean distance D (x) from the vessel wall from the starting point.

Using the level set fast marching method of Equation 2, the time it takes to propagate the level set to each point is quickly calculated.

Points having the same integer value of the time taken for the level set to be propagated are defined as one set, and points corresponding to the same set are connected in the same color pattern as shown in FIG. 5.

Here, T _{i, j} represents the time taken for a specific level set wave to propagate from the starting point to the two-dimensional pixel position (i, j).

And the set of points with only one adjacent set becomes the distal end of the vessel. That is, the end of a set of points that has data for the previous point and no data for the next point. Eventually, the center point of the distal end can be extracted as the end point of the blood vessel. Vascular skeletalization is achieved by backtracking the parent nodes stored in the path tree from each of the extracted end points to the starting point. When the skeleton is already extracted, the location is marked as a branch point and the backtracking is stopped.

Figure 6 is an exemplary view showing the extraction results of hepatic portal skeleton, branch points, endpoints according to an embodiment of the present invention. Since the path tree is propagated from the points far from the blood vessel wall, the traceback of the path tree eventually follows the points far from the blood vessel wall, so that the skeleton of the blood vessel can be extracted as shown in FIG.

Next, the branch point position correction unit 140 correctly corrects the position of the extracted branch point (S240).

The branching points of blood vessels are created where the centerline and the centerline meet. In this case, the shape of the hepatic context is not a straight form, but a reason for the excessive curvature, such that a part to be recognized as one branch point may be recognized as several branch points. Therefore, according to the embodiment of the present invention, the accuracy of the branch point extraction is improved by adding the extracted branch point position correction step.

7 is a view for explaining a branch point position correction process according to an embodiment of the present invention.

According to the exemplary embodiment of the present invention, the maximum sphere inscribed in the vessel wall is calculated by expanding the virtual sphere around the points in the skeleton adjacent to the predetermined segment with respect to the branch points extracted in step S230.

As a result, the maximum radius is calculated for each point, which can be assumed to be a local approximate vessel thickness. Therefore, the branch point position can be corrected to a point (point b) in the skeleton close to the initial branch point (point a) having the maximum radius r. In other words, the point corresponding to the center of the largest sphere is the position of the corrected branch point.

More specifically, the process of finding the point where the vascular wall and the surface of the virtual sphere meet by expanding the virtual sphere having a radius of 1 voxel by 1 voxel centered on the point P _candidate in the skeleton near the first calculated branch point. Done.

In order to more quickly correct the position of the branch point, before creating the actual skeleton, an offset table is stored in which relative coordinates are stored in order of close distance from the center of the sphere to the surface of the virtual sphere. Therefore, the absolute coordinates of the P _candidate corresponding to the center of the virtual sphere change at each branch point correction step, but when the relative coordinates stored in the offset table (OT) are added to the coordinates of this center, The coordinate P _sphere of the imaginary sphere surface is quickly calculated.

Here, P _candidate represents a coordinate (x ₀ , y ₀ , z ₀ ) corresponding to the center of the virtual sphere.

In the offset table, the distance is increased by 1 voxel from the points corresponding to the surface of the sphere with a radius of 1 voxel based on the point (x ₀ , y ₀ , z ₀ ) corresponding to the center of the sphere. Store the coordinates in sequence. Increasing the distance value corresponds to inflating the virtual sphere. In the two-dimensional example, because of symmetry, only the coordinates of the first quadrant can be considered.

의 올림 값을 구하여 D 라고 하면, D 값이 같은 점들을 도 8과 같이 분류한다. That is, the Euclidean distance from the center of the point (x, y) with all integer values around the point of (x ₀ , y ₀ )

If the rounding value is obtained and is D, the points having the same D value are classified as shown in FIG.

8 is a diagram illustrating two-dimensional coordinate classification according to a distance D in a first quadrant according to an embodiment of the present invention. 8 is a result calculated only for the first quadrant, so the x-axis, y-axis, and origin-symmetric values are stored in the offset table starting from the coordinates of the point having a small distance from the center. This data structure speeds up the computational process of inflating a virtual sphere during automatic path finding.

  Next, the branch pattern analysis unit 150 analyzes the branching pattern with respect to the liver context and extracts the main blood vessel structure (S250). First, a branch point located at the entrance of the liver context may be set as a parent node, and the level of each sub-branch point may be calculated as shown in FIG. 9 through vascular structure analysis.

9 is a view for explaining the analyzed level of each sub-branch according to an embodiment of the present invention. As shown in FIG. 9, a parent node corresponding to an upper node corresponds to a first level, and nodes corresponding to the lower node may be represented as a second level, a third level, or the like.

In this way, by setting the level for the branching point, the branching pattern for the tree-shaped liver context can be accurately analyzed using the level information and the skeleton information.

The diameter of the hepatic portal vein is estimated approximately at a specific location on the skeleton using Euclidean distance map in three dimensions generated from the hepatic portal vascular boundary. By calculating the average diameter of each of the detail vessels, the secondary vessels are removed using a diameter threshold that distinguishes the primary vessels and the secondary vessels.

  Finally, the liver segment separator 160 automatically classifies the liver segment according to the branching pattern structure of the main blood vessels of the liver context (S260). According to an exemplary embodiment of the present invention, as shown in FIGS. 10 and 11, liver segments may be divided by reflecting anatomical clinical information on branch patterns of liver contexts different for each person.

Figure 10 shows the anatomical structure of the liver left segment according to an embodiment of the present invention, Figure 11 shows the anatomy of the liver right segment according to an embodiment of the present invention.

  The left segment of the liver may be divided into a general case as shown in (a) of FIG. 10 and a case where more than two sub-branches exist as shown in (b) of FIG. 10.

From the skeleton information of the hepatic context, after determining whether the actual patient's case belongs to either, according to FIG. 10, three kinds of S2 (lateral, superior, posterior), S3 (lateral, inferior, anterior), and S4 (medial) Separate into segments.

In addition, the right segment of the liver has a general case as shown in FIG. 11A and multiple posterior branches as shown in FIG. 11B, trifurcation as shown in FIG. 11C. In the case of, can be divided into the case of early branching as shown in (d) of FIG.

As described above, the branch pattern of the left liver segment is divided into two types as shown in FIG. 10, and the branch pattern of the liver right segment is divided into four types as shown in FIG. 11.

Accordingly, according to an embodiment of the present invention, it is possible to determine which branch pattern of the liver context of a specific person belongs to the pattern shown in FIGS. 10 and 11 by using the end point, branch point, and skeletal extraction result of the liver context. The pattern can be classified accordingly.

12 is a view for explaining the two-dimensional projection positional relationship of liver right segments according to an embodiment of the present invention. Based on the skeleton structure information of the hepatic context and the positional relationship in the two-dimensional projection image of FIG. 11, after determining whether the actual patient's case belongs to either, RA (anterior, lateral, superior), RP (posterior, medial, inferior).

Here, the calculation of which anatomical liver segments belong to a particular voxel inside the liver may be performed by calculating whether the particular voxel is close to the Eucladian distance of the branch of the liver context of the segment.

FIG. 13 is an exemplary diagram illustrating a distance-based liver segment classification modeling method according to a blood vessel structure of a liver context according to an embodiment of the present invention.

For example, assuming that there are four branches of the liver context and blood is supplied from the nearest branch as shown in FIG. 13, the Euclidean distance is closest to the voxel inside the liver as shown in Equation 4 below. Can compute branch

Herein, the v'divided voxel inside the liver, and the v 'represents the blood vessel corresponding to the i-th branch among the four branches of the context, d _i (v) means the minimum distance between the i-th branch and the voxel v.

That is, as shown in FIG. 13, there are two branches at the first level of the liver context with respect to the voxel v inside the liver, and two branches each come out of the branches at the second level. Therefore, d ₁ (v) is the minimum distance between the first branch of the four branches and the voxel v, and d ₂ (v) is the minimum distance between the second branch of the four branches and the voxel v. d ₃ (v) is the minimum distance between the third branch of the four branches and the voxel v, and d ₄ (v) is the minimum distance between the fourth branch of the four branches and the voxel v.

If we find the minimum value d _K (v) of these four distances, we can say that the particular voxel v is closest to the vessel used in the distance calculation and that blood is supplied from the branch.

14 and 15 illustrate a result of applying an automatic liver classification method using vascular structure information of the liver context to an abdominal CT image according to an embodiment of the present invention.

14 is an exemplary diagram showing a segment classification result of a left liver segment to which the present invention is applied. (A) of FIG. 14 shows that it is divided into one segment of S2, (b) shows that it is divided into two segments of S2 and S3, and (c) is divided into three segments of S2, S3 and S4. Indicates that Here, S2 represents lateral, superior, posterior, S3 represents lateral, inferior, and anterior, and S4 represents medial.

15 is an exemplary diagram showing a segment classification result of a liver right segment to which the present invention is applied.

In FIG. 15, two segments, RA, anterior, lateral, and superior, and RP (posterior, medial, inferior) of the right hepatic segment are well distinguished.

While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

100: intersegment separator, 110: volume data acquisition unit,
120: path tree generation unit, 130: skeleton extracting unit,
140: branch point position correction unit, 150: branch pattern analysis unit,
160: intersegment separator

Claims (12)

Acquiring three-dimensional chest volume data through a CT scan;
Generating a path tree using a Eucladian distance map of points at blood vessel boundaries of the liver context using the 3D chest volume data;
Extracting end points, branches and skeletons of the hepatic context using a level set,
Correcting the position of the extracted branch points,
Analyzing the branching pattern for the liver context and extracting the main blood vessel from the skeleton, and
The liver segment classification method using the vascular structure information of the liver context comprising the step of classifying the liver segment using the branched pattern structure of the main blood vessel. The method of claim 1,
Generating the path tree,
Generating the eucladian distance map for all points at the vessel border of the hepatic portal context, and
Selectively connecting the inner voxels farthest from the vascular boundary to generate the pathway tree. The method of claim 2,
Extracting the end point, branch point and skeleton of the liver context,
Defining points having the same integer value as the time taken for the level set to propagate as a set, and determining a set of points having one adjacent set as an end point of the intercontext;
Extracting skeletal information of the liver context by connecting back traced from the end point of the extracted liver context to a starting point, and
The liver segment classification method using the blood vessel structure information of the liver context comprising the step of judging the intersection point of the extracted plurality of the skeleton as a branch point. The method of claim 3,
Correcting the position of the extracted branch point,
Calculating a sphere having the largest size inscribed in the blood vessel wall of the hepatic portal vein by expanding the virtual sphere around the candidate points adjacent to the extracted branch point; and
And correcting the position of the branch point to a point corresponding to the center point of the sphere of the largest size. 5. The method of claim 4,
As shown in the following equation, the coordinates of the virtual sphere are added by adding the relative coordinates OT _{x (i)} , OT _{y (i)} , and OT _{z (i)} stored in the offset table to the center coordinates (P _candidate ) for the candidate points. To classify liver segments using vessel structure information of the liver context to calculate (P _sphere ):

Here, P _candidate represents a coordinate (x ₀ , y ₀ , z ₀ ) corresponding to the center of the virtual sphere. The method of claim 5,
Analyzing the branching pattern for the liver context and extracting the main blood vessel structure,
Setting a branch point located at an entrance of the liver context as a parent node, and setting a level for the parent node and branch points connected to the parent node,
Analyzing a branching pattern for the liver context in a tree form from the level information and skeleton information,
Predicting the diameter of the hepatic context via a Eucladian distance map for the points at the boundary of the hepatic context, and
And dividing the main blood vessel and the secondary blood vessel using the diameter and the threshold value of the hepatic portal vein, and removing the secondary blood vessel. Volume data acquisition unit for acquiring three-dimensional chest volume data through a CT scan,
A route tree generator for generating a route tree using a Eucladian distance map of points at a blood vessel boundary of a liver context using the 3D chest volume data;
Skeleton extracting unit for extracting the end point, branch point and the skeleton of the liver context using a level set,
Branch point position correction unit for correcting the position of the extracted branch point,
Branch pattern analysis unit for analyzing the branching pattern for the liver context and extracting the main blood vessels from the skeleton, and
Liver segment classification device using the vascular structure information of the liver context comprising a liver segment separator for classifying the liver segment using the branch pattern structure of the main blood vessel. The method of claim 7, wherein
The path tree generation unit,
Vascular structure of the hepatic portal vein, generating the Eucladian distance map for all points at the vascular border of the hepatic portal vein, and selectively connecting the inner voxels farthest from the vascular border of the hepatic portal portal to generate the path tree Device for distinguishing between segments using information. 9. The method of claim 8,
The skeleton extracting unit,
Defines points having the same integer value as the time taken for the level set to propagate as a set, determines a set of points having an adjacent set as the end point of the intercontext, and starts from the end point of the extracted hepatic context. A device for separating liver segments using blood vessel structure information of the liver context, which traces back to and connects to extract skeleton information of the liver context, and determines a point where the extracted plurality of skeletons intersect as a branch point. 10. The method of claim 9,
The branch point position correction unit,
The virtual sphere is expanded around the candidate points adjacent to the extracted branch point to calculate a sphere having the largest size inscribed to the vessel wall of the hepatic portal vein, and the branch point is a point corresponding to the center point of the sphere having the maximum size. A device for distinguishing liver segments using blood vessel structure information of the liver context to correct the position of the liver. The method of claim 10,
The branch point position correction unit,
As shown in the following equation, the coordinates of the virtual sphere are added by adding the relative coordinates OT _{x (i)} , OT _{y (i)} , and OT _{z (i)} stored in the offset table to the center coordinates (P _candidate ) for the candidate points. Liver segment classification device using vascular structure information of liver context to calculate (P _sphere ).

Here, P _candidate represents a coordinate (x ₀ , y ₀ , z ₀ ) corresponding to the center of the virtual sphere. 12. The method of claim 11,
The branch pattern analysis unit,
A branch point located at the entrance of the liver context is set as a parent node, a level is set for the parent node and branch points connected to the parent node, and the inter-context in the form of a tree from the level information and skeleton information Analyze the branching pattern for, predict the diameter of the hepatic portal vein using a Eucladian distance map of the points at the border of the hepatic portal vein, and use the diameter and the threshold of the hepatic portal vein to use the main and secondary vessels. A device for classifying liver segments using blood vessel structure information of the liver context for classifying and removing the secondary blood vessels.

Images