Method and system for segmenting artery region in liver medical images

Abstract

The invention discloses a method and system for segmenting arterial regions of a liver medical image. The method comprises: removing muscle tissue and liver tissue on a liver CT image; selecting a data space on a liver blood vessel contour as an operation region; calculating a Hessian matrix function value of each point in the operation region to obtain a function value set, selecting a specified step length in the function value set to extract a seed point; calculating a normal curvature of each obtained seed point, setting two principal curvatures of the seed point, referring to the parameter range of the two principal curvatures that have been set, and retaining seed points whose principal curvatures are within the parameter range; tracking the retained seed points to obtain a center line of the blood vessel, comparing the grayscale values ​​of pixels perpendicular to a tracking direction of a topological thinning method, setting a grayscale value of a blood vessel boundary to obtain a blood vessel boundary; and determining the number of blood vessels and a blood vessel radius, setting a major axis and a minor axis of the blood vessel according to the blood vessel radius, using an ellipse to depict the cross-sectional shape of the blood vessel for modeling, and obtaining a liver artery segmentation image.

Description

Technical Field

[0001] The invention relates to a medical image processing system, and in particular to a liver medical image processing system. Background Art

[0002] Usually, patients with liver disease need to collect liver CT angiography images before surgery to show the anatomical structure of the liver artery, locate the liver artery that supplies blood to the lesion, and minimize the impact of the surgery on the liver tissue. At present, the segmentation of the arterial area is estimated based on the doctor's clinical judgment and naked eye observation, which often affects the surgical effect or causes the failure of the operation due to inaccuracy. Therefore, the medical field has been working hard to develop a system that can automatically segment the liver artery area.

[0003] For example, a method for segmenting the hepatic artery based on CT images disclosed in Chinese patent application No. 201510313660.X includes: loading the hepatic artery CT image data and the corresponding liver mask data, and locating the artery seed point; segmenting the CT image data to obtain first binary volume data containing the spine and its connected ribs and the hepatic artery, and second binary volume data containing the spine and its connected ribs and the hepatic artery; removing the same part as the first binary volume data from the second binary volume data to obtain the third binary volume data; using the trunk artery seed point to perform three-dimensional region growth on the third binary volume data to obtain the hepatic artery segmentation result. However, the segmentation method lacks fine processing for the situation where there are blood vessels crossing or attached to other tissues in the segmented area, resulting in low accuracy of the obtained hepatic artery segmentation image, which is not conducive to lesion identification.

[0004] Another example is a coronary artery segmentation method and device disclosed in Chinese patent application No. 201410132571.0, which includes the following steps: (1) determining the position of the aorta in the original image, and performing enhancement filtering on the original image to obtain an enhanced filtered image; (2) obtaining the starting areas of the left and right coronary arteries based on the position of the aorta, the original image, and the grayscale of the enhanced filtered image; (3) segmenting the enhanced rate wave image based on at least one preset segmentation threshold, obtaining the candidate area corresponding to the preset threshold, and determining the candidate area of ​​the left and right coronary arteries based on the candidate area and the starting areas of the left and right coronary arteries; (4) screening the left and right coronary artery candidate areas respectively according to the positional relationship between the aorta and the left and right artery candidate areas, and connecting them with the starting areas after screening to obtain the left and right coronary arteries. However, the coronary arteries segmented by this method have the problem of blood vessels crossing or adhering to other tissues.

[0005] Therefore, it is an urgent problem to provide a method and system for segmenting the arterial region of liver medical images that can quickly, efficiently and accurately segment the liver artery vessels in liver medical images. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for segmenting the arterial region of a liver medical image. The method is fast and efficient, solves the problem of blood vessels crossing or adhering to other tissues, and the liver artery segmentation image obtained after segmentation processing has high accuracy.

[0007] In order to achieve the above-mentioned object, one aspect of the present invention provides a method for segmenting an artery region of a liver medical image, comprising: (1) removing muscle tissue and liver tissue on a liver CT image to expose the liver vascular contour; (2) selecting two diagonal points on the liver vascular contour that are the farthest apart, establishing a cuboid with the selected two points as diagonals, and using the data space contained in the cuboid as an operating area; (3) calculating the Hessian matrix function value of each point in the operating area through a Hessian matrix function to obtain a function value set, and selecting a specified step size in the function value set to extract a seed point, wherein the specified step size is set to be less than 5 voxels; (4) calculating the normal curvature of each seed point obtained in step (3), The maximum and minimum values ​​of the normal curvature are set as the two principal curvatures of the seed point, and the parameter range of the two principal curvatures is referred to, and the seed points whose principal curvatures are not within the parameter range are eliminated to eliminate the non-tubular tissue in the liver vascular contour, and the seed points whose principal curvatures are within the parameter range are retained; (5) based on the seed points retained in the topological refinement method tracking step (4), the center line of the blood vessel is obtained, the gray value of the pixel perpendicular to the tracking direction of the topological refinement method is compared, and the gray value of the blood vessel boundary is set to obtain the blood vessel boundary; and (6) the number of blood vessels is determined, the blood vessel radius is calculated according to the distance from the center line to the blood vessel boundary, the major axis and minor axis of the blood vessel are set according to the blood vessel radius, and the blood vessel cross-sectional shape is described by an ellipse to model the model, and the liver artery segmentation image is obtained.

[0008] Optionally, in step (5), the quotient of the two principal curvatures is both 0.

[0009] Optionally, the method further comprises, before step (1), performing enhancement processing on the liver CT image.

[0010] Among them, the Hessian matrix, usually called the Hessian matrix, is also translated as the Hessian matrix, the Hessian matrix, the Hessian matrix, etc. It is a square matrix composed of the second-order partial derivatives of a multivariate function, which describes the local curvature of the function. The Hessian matrix was first proposed by the German mathematician Ludwig Otto Hesse in the 19th century and named after him. In the optimization design of practical engineering problems, the objective function is often expanded into a Taylor polynomial in the neighborhood of a certain point to approximate the original function. At this time, the Hessian matrix is ​​involved in the matrix form of the Taylor expansion of the function at a certain point.

[0011] Another aspect of the present invention is to provide a liver medical image artery region segmentation system, comprising: a data reading module, which is used to read a liver CT image; an operation region division module, which is connected to the data reading module in communication, and is used to select two diagonal points with the farthest distance on the liver CT image, and to establish a cuboid with the selected two points as the diagonal, and to use the data space contained in the cuboid as the operation region; a function calculation module, which is connected to the operation region division module in communication, and is used to calculate the Hessian matrix function value of each point in the operation region through the Hessian matrix function, obtain a function value set, and select a specified step length in the function value set to extract a seed point, wherein the specified step length is set to be less than 5 voxels; a seed point confirmation module, which is connected to the function calculation module in communication, and is used to calculate the function calculation module. The block obtains the normal curvature of each seed point, sets the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point, sets the parameter range with the two principal curvatures as the upper and lower limits, removes the seed points whose principal curvatures are not within the parameter range, and retains the seed points whose principal curvatures are within the parameter range; a centerline extraction module, which is communicated with the seed point confirmation module, is used to track the seed points retained by the seed point confirmation module based on the topological refinement method, obtain the centerline of the blood vessel, compare the grayscale values ​​of the pixels perpendicular to the tracking direction of the topological refinement method, and set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary; and a three-dimensional drawing module, which is communicated with the centerline extraction module, is used to calculate the blood vessel radius according to the distance from the centerline to the blood vessel boundary after determining the number of blood vessels, set the major axis and minor axis of the blood vessel according to the blood vessel radius, and use an ellipse to depict the cross-sectional shape of the blood vessel for modeling.

[0012] Optionally, the method further comprises an image preprocessing module which is communicatively connected to the data reading module and the operation area division module, and the image preprocessing module comprises an image smoothing unit and an image enhancement unit which are communicatively connected to each other, wherein the image smoothing unit is used to directly perform operation processing on the pixel grayscale value of each liver CT image in the spatial domain by a spatial domain method to filter out noise in each liver CT image; and the image enhancement unit is used to sharpen and enhance each liver CT image processed by the image smoothing processing unit.

[0013] Optionally, the seed point confirmation module is further used to calculate a shape index of the arterial blood vessel.

[0014] Optionally, the seed point confirmation module includes a normal curvature calculation unit and a seed point retention unit that are communicatively connected, wherein the normal curvature calculation unit is used to calculate the normal curvature of each seed point obtained by the function calculation module, and set the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point; the seed point retention unit is used to refer to the parameter range of the two set principal curvatures, eliminate seed points whose principal curvatures are not within the parameter range, and retain seed points whose principal curvatures are within the parameter range.

[0015] Optionally, the centerline extraction module includes a center point detection unit and a blood vessel boundary detection unit that are communicatively connected, wherein the center point detection unit is used to track the seed points retained by the seed point confirmation module based on the topological refinement method to obtain the centerline of the blood vessel; the blood vessel boundary detection unit is used to compare the grayscale values ​​of pixels perpendicular to the tracking direction of the topological refinement method, and set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary.

[0016] Optionally, the three-dimensional drawing module includes: a communication-connected blood vessel number calculation unit, a blood vessel radius measurement unit, and a region merging and drawing unit, wherein the blood vessel number calculation unit is used to perform region cropping on the liver CT image to determine the number of blood vessels; the blood vessel radius measurement unit is used to measure the blood vessel radius based on the distance from the center line to the blood vessel boundary; the region merging and drawing unit is used to set the major axis and minor axis of the blood vessel according to the blood vessel radius, and use an ellipse to depict the cross-sectional shape of the blood vessel for modeling.

[0017] Optionally, a liver artery database is further included that is communicatively connected to the three-dimensional rendering module and is used to store liver source information, a liver CT image of the corresponding liver, and a corresponding liver artery segmentation image.

[0018] Optionally, the liver source information includes at least source information such as the disease type, patient gender, patient age, living area, and hospital visited of the uploaded corresponding liver CT image.

[0019] Optionally, the liver source information may also include: patient living habits, biochemical examination information, patient typical symptoms, signs, medical images, imaging diagnosis results, treatment plans, adverse reactions and attending physicians, etc.

[0020] Optionally, the method further comprises a retrieval module connected to the database for searching the liver CT image of the target liver and the corresponding liver artery segmentation image in the liver artery database.

[0021] Optionally, the parameters set by the function calculation module may also include: arterial blood vessel Gaussian curvature, mean curvature, tortuosity and other related parameters.

[0022] Alternatively, both the principal curvature and the shape index can describe the different shapes of various objects. The quotient of the two principal curvatures of an ideal tubular structure is approximately 0, and the shape index is approximately equal to 0.75. The shape index of a blood vessel ranges from 0.65 to 0.8.

[0023] Optionally, the parameter range of the principal curvature may be 0.5 to 0.7.

[0024] Optionally, the blood vessel number calculation unit sets that each blood vessel on the liver CT image has one and only one bright area, that is, each blood vessel passes through each slice only once, the blood vessel changes direction smoothly, and all blood vessels start at the first slice, that is, no new nerve blood vessels appear in the remaining slices.

[0025] The beneficial effects of the present invention are as follows: (1) the present invention establishes an elliptical blood vessel model using a center line, a major axis and a minor axis. Since each blood vessel is not uniform, different parts of different blood vessels have different radii, but for each blood vessel, the change in radius is continuous, avoiding errors caused by setting the blood vessel as a circle, and making the obtained segmented image closer to the real blood vessel; (2) the present invention removes non-tubular tissues in liver CT images, and the three-dimensional drawing module further determines the number of arterial blood vessels, avoiding the loss of the extracted center line due to attachment or intersection of blood vessels in three-dimensional space, which affects the segmentation accuracy; (3) the segmentation system of the present invention sets blood vessel parameters, such as principal curvature and shape index, through a function calculation module, which can depict small shape changes such as fluctuations and concavities of the arterial blood vessel surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the structure of the liver medical image artery region segmentation system of the present invention.

[0027] Figure 2 It is a schematic diagram of the flow of the liver medical image artery region segmentation method of the present invention. DETAILED DESCRIPTION

[0028] Please refer to Figure 1 According to a non-limiting embodiment of the present invention, a liver medical image artery region segmentation system is provided, comprising: a data reading module 10, an operation region division module 20, a function calculation module 30, a seed point confirmation module 40, a centerline extraction module 50, and a three-dimensional rendering module 60.

[0029] The data reading module 10 is used to read the liver CT image.

[0030] The operation region division module 20 is used to select two diagonal points with the farthest distance from each other on the liver CT image, establish a cuboid with the two selected points as the diagonal line, and use the data space contained in the cuboid as the operation region.

[0031] The function calculation module 30 is used to calculate the Hessian matrix function value of each point in the operating area through the Hessian matrix function, obtain a function value set, and select a specified step size in the function value set to extract a seed point, wherein the specified step size is set to less than 5 voxels.

[0032] The seed point confirmation module 40 is used to calculate the normal curvature of each seed point obtained by the function calculation module, set the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point, refer to the parameter range of the two principal curvatures that have been set, eliminate the seed points whose principal curvatures are not within the parameter range, and retain the seed points whose principal curvatures are within the parameter range.

[0033] The centerline extraction module 50 is used to track the seed points retained by the seed point confirmation module based on the topology thinning method to obtain the centerline of the blood vessel, compare the grayscale values ​​of pixels perpendicular to the tracking direction of the topology thinning method, and set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary.

[0034] The three-dimensional drawing module 60 is used to measure the vessel radius according to the distance from the center line to the vessel boundary after determining the number of blood vessels, set the long axis and short axis of the vessel according to the vessel radius, and use an ellipse to describe the cross-sectional shape of the blood vessel for modeling.

[0035] As an alternative embodiment, an image preprocessing module 70 is further included which is in communication with the data reading module 10 and the operation region division module 20. The image preprocessing module 70 includes an image smoothing unit 710 and an image enhancement unit 720. The image smoothing unit 710 is used to directly perform operation processing on the pixel grayscale value of each liver CT image in the spatial domain by using a spatial domain method, thereby filtering out noise in each liver CT image. The image enhancement unit 720 is used to sharpen and enhance each liver CT image processed by the image smoothing processing unit.

[0036] As another alternative implementation, the seed point confirmation module 40 includes a normal curvature calculation unit 410 and a seed point retention unit 420 that are communicatively connected. The normal curvature calculation unit 410 is used to calculate the normal curvature of each seed point obtained by the function calculation module, and set the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point. The seed point retention unit 420 is used to refer to the parameter range of the two principal curvatures that have been set, eliminate the seed points whose principal curvatures are not within the parameter range, and retain the seed points whose principal curvatures are within the parameter range.

[0037] The centerline extraction module 50 includes a center point detection unit 510 and a blood vessel boundary detection unit 520. The center point detection unit 510 is used to track the seed points retained by the seed point confirmation module based on the topology thinning method to obtain the centerline of the blood vessel. The blood vessel boundary detection unit 520 is used to compare the grayscale values ​​of pixels perpendicular to the tracking direction of the topology thinning method, set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary.

[0038] The three-dimensional drawing module 60 includes a blood vessel number calculation unit 610, a blood vessel radius measurement unit 620, and a region merging drawing unit 630. The blood vessel number calculation unit 610 is used to perform region cropping on the liver CT image to determine the number of blood vessels. The blood vessel radius measurement unit 620 is used to measure the blood vessel radius according to the distance from the center line to the blood vessel boundary. The region merging drawing unit 630 is used to set the long axis and short axis of the blood vessel according to the blood vessel radius, and use an ellipse to depict the cross-sectional shape of the blood vessel for modeling.

[0039] As another alternative embodiment, the liver medical image artery region segmentation system further includes a liver artery database 80 and a retrieval module 90. The liver artery database 80 is used to store liver source information, a liver CT image of a corresponding liver, and a corresponding liver artery segmentation image. The retrieval module 90 is used to retrieve a liver CT image of a target liver and a corresponding liver artery segmentation image from the liver artery database 80.

[0040] Next, combine Figure 2 , the liver medical image artery region segmentation method of the present invention is described in detail. Figure 2 As shown, first, in step S1, the muscle tissue and liver tissue on the liver CT image are removed to expose the liver vascular contour. Then, in step S2, two diagonal points with the farthest distance on the liver vascular contour are selected, and a cuboid is established with the selected two points as diagonals, and the data space contained in the cuboid is used as the operation area. Then, in step S3, the Hessian matrix function value of each point in the operation area is calculated by the Hessian matrix function to obtain a function value set, and a specified step size is selected in the function value set to extract the seed point, wherein the specified step size is set to be less than 5 voxels. Subsequently, in step S4, the normal curvature of each seed point obtained in step S3 is calculated, and the maximum and minimum values ​​of the normal curvature are set as the two principal curvatures of the seed point. Referring to the parameter range of the two principal curvatures that have been set, the seed points whose principal curvatures are not within the parameter range are eliminated to eliminate the non-tubular tissue in the liver vascular contour, and the seed points whose principal curvatures are within the parameter range are retained; in step S5, the seed points retained in step S4 are tracked based on the topological thinning method to obtain the center line of the blood vessel, and the grayscale values ​​of the pixels perpendicular to the tracking direction of the topological thinning method are compared, and the grayscale value of the blood vessel boundary is set to obtain the blood vessel boundary. Finally, in step S6, the number of blood vessels is determined, the blood vessel radius is calculated according to the distance from the center line to the blood vessel boundary, the major axis and minor axis of the blood vessel are set according to the blood vessel radius, and the cross-sectional shape of the blood vessel is depicted and modeled using an ellipse.

[0041] Although the preferred embodiments of the present invention have been described in detail herein, it is to be understood that the present invention is not limited to the specific structures and steps described and shown in detail herein, and other variations and modifications may be implemented by those skilled in the art without departing from the spirit and scope of the present invention.

Claims

1. A method for segmenting arterial regions of a liver medical image, comprising: (1) Remove the muscle tissue and liver tissue on the liver CT image to expose the liver vascular contour; (2) selecting two diagonal points on the contour of the liver blood vessels that are the farthest apart, establishing a cuboid with the two selected points as diagonals, and using the data space contained in the cuboid as the operating area; (3) calculating the Hessian matrix function value of each point in the running area through the Hessian matrix function to obtain a function value set, and selecting a specified step size from the function value set to extract a seed point, wherein the specified step size is set to be less than 5 voxels; (4) calculating the normal curvature of each seed point obtained in step (3), setting the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point, referring to the parameter range of the two principal curvatures that have been set, eliminating the seed points whose principal curvatures are not within the parameter range, so as to eliminate the non-tubular tissue in the liver vascular contour, and retaining the seed points whose principal curvatures are within the parameter range; (5) based on the topological thinning method, tracking the seed points retained in step (4), obtaining the center line of the blood vessel, comparing the gray values ​​of pixels perpendicular to the tracking direction of the topological thinning method, setting the gray value of the blood vessel boundary to obtain the blood vessel boundary; and (6) Determine the number of blood vessels, calculate the blood vessel radius according to the distance from the center line to the blood vessel boundary, set the major axis and minor axis of the blood vessel according to the blood vessel radius, use an ellipse to describe the cross-sectional shape of the blood vessel and model it, and obtain a liver artery segmentation image.

2. The method for segmenting liver medical image artery regions according to claim 1, characterized in that: The method further includes performing enhancement processing on the liver CT image before step (1).

3. A liver medical image artery region segmentation system, characterized in that: include: A data reading module, which is used to read liver CT images; An operation area division module, which is in communication connection with the data reading module, is used to select two diagonal points with the farthest distance from each other on the liver CT image, establish a cuboid with the selected two points as diagonals, and use the data space contained in the cuboid as the operation area; A function calculation module, which is in communication with the operation area division module, is used to calculate the Hessian matrix function value of each point in the operation area through a Hessian matrix function, obtain a function value set, and select a specified step size from the function value set to extract a seed point, wherein the specified step size is set to be less than 5 voxels; A seed point confirmation module, which is in communication with the function calculation module and is used to calculate the normal curvature of each seed point obtained by the function calculation module, set the maximum and minimum values ​​of the normal curvature as the two principal curvatures of the seed point, and refer to the parameter range of the two principal curvatures that have been set, remove the seed points whose principal curvatures are not within the parameter range, and retain the seed points whose principal curvatures are within the parameter range; a centerline extraction module, which is in communication with the seed point confirmation module, and is used to track the seed points retained by the seed point confirmation module based on the topology thinning method to obtain the centerline of the blood vessel, compare the grayscale values ​​of pixels perpendicular to the tracking direction of the topology thinning method, and set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary; and The three-dimensional drawing module is in communication with the centerline extraction module and is used to calculate the blood vessel radius according to the distance from the centerline to the blood vessel boundary after determining the number of blood vessels, set the major axis and minor axis of the blood vessel according to the blood vessel radius, and use an ellipse to depict the cross-sectional shape of the blood vessel for modeling.

4. The liver medical image artery region segmentation system according to claim 3, characterized in that: It further comprises an image preprocessing module which is in communication connection with the data reading module and the operation area division module, wherein the image preprocessing module comprises an image smoothing unit and an image enhancement unit which are in communication connection with each other, The image smoothing unit is used to directly perform calculation processing on the pixel grayscale value of each liver CT image in the spatial domain by using a spatial domain method to filter out noise in each liver CT image; The image enhancement unit is used for sharpening and enhancing each liver CT image processed by the image smoothing unit.

5. The liver medical image artery region segmentation system according to claim 3, characterized in that: The seed point confirmation module is also used to calculate the shape index of the arterial blood vessel.

6. The liver medical image artery region segmentation system according to claim 3, characterized in that: The seed point confirmation module includes a normal curvature calculation unit and a seed point reservation unit connected in communication. The normal curvature calculation unit is used to calculate the normal curvature of each seed point obtained by the function calculation module, and set the maximum value and the minimum value of the normal curvature as the two principal curvatures of the seed point; The seed point retaining unit is used to refer to the parameter ranges of the two set principal curvatures, remove seed points whose principal curvatures are not within the parameter ranges, and retain seed points whose principal curvatures are within the parameter ranges.

7. The liver medical image artery region segmentation system according to claim 3, characterized in that: The centerline extraction module includes a center point detection unit and a blood vessel boundary detection unit connected in communication. Wherein, the center point detection unit is used to track the seed points retained by the seed point confirmation module based on the topological refinement method to obtain the center line of the blood vessel; The blood vessel boundary detection unit is used to compare the grayscale values ​​of pixels perpendicular to the tracking direction of the topology thinning method, and set the grayscale value of the blood vessel boundary to obtain the blood vessel boundary.

8. The liver medical image artery region segmentation system according to claim 7, characterized in that: The three-dimensional rendering module includes: a blood vessel number calculation unit, a blood vessel radius measurement unit, and a region merging rendering unit connected to each other. Wherein, the blood vessel number calculation unit is used to perform regional cropping on the liver CT image to determine the number of blood vessels; The blood vessel radius measurement unit is used to measure the blood vessel radius according to the distance from the center line to the blood vessel boundary; The region merging and drawing unit is used to set the major axis and the minor axis of the blood vessel according to the blood vessel radius, and to depict the cross-sectional shape of the blood vessel by using an ellipse for modeling.

9. The liver medical image artery region segmentation system according to claim 3, characterized in that: It also includes a liver artery database that is communicatively connected to the three-dimensional rendering module and is used to store liver source information, a liver CT image of the corresponding liver, and a corresponding liver artery segmentation image.

10. The liver medical image artery region segmentation system according to claim 9, characterized in that: It also includes a retrieval module connected to the database for searching the liver CT image of the target liver and the corresponding liver artery segmentation image in the liver artery database.

Images