Method for Extracting Liver Segmentation Boundary Based on Bone Information

Abstract

The present invention relates to a method for extracting the liver segmentation boundary based on bone information, comprising: extracting bone information in a CT image through image processing; using morphological operations and anatomical knowledge to denoise and completely segment the bone; using statistical methods to extract the longitudinal boundary of liver segmentation according to the bone information; and using image processing techniques to extract the transverse boundary of liver segmentation from the bone information. The present invention helps to assist the liver segmentation algorithm, can be used to reduce the misrecognition rate of the liver area outside the abdominal wall, and reduce the processing time of the liver segmentation algorithm.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and specifically relates to a method for extracting the liver segmentation boundary based on bone information. Background Art

[0002] Due to the complexity of the liver's anatomical structure, the low contrast between the liver and adjacent organs, and the presence of lesions, the accurate segmentation of the liver region remains a difficult task. For example, since the CT values of the skin, muscle, soft tissue, and liver are similar, incorrect identification beyond the sternum range may occur when segmenting the liver region. Due to the large number of single CT slices and the large number of pixels in a single slice image, the processing time of the liver segmentation algorithm is relatively long.

[0003] Currently, the liver segmentation problem on CT images takes a long time and is easily interfered by other tissues. The present invention proposes a solution to the problem of the liver segmentation boundary in CT images. Summary of the Invention

[0004] The present invention aims to overcome the above-mentioned disadvantages of the prior art and provides a method for extracting the liver segmentation boundary based on bone information.

[0005] To solve the above technical problems, the present invention proposes a method for extracting the liver segmentation boundary based on bone information, including the following steps:

[0006] S110, extracting the bone information in the CT image through image processing;

[0007] S120, using morphological operations and anatomical knowledge to denoise and completely segment the bones;

[0008] S130, using statistical methods to extract the longitudinal boundary of liver segmentation based on the bone information;

[0009] S140, using image processing technology to extract the transverse boundary of liver segmentation from the bone information.

[0010] Further, the extracting the bone information in the CT image through image processing in step S110 specifically includes:

[0011] S1101 performing image preprocessing to map the CT value range from -1000 to 1000 to the pixel interval from 0 to 255;

[0012] S1102 using a high threshold to remove high-density noise, such as metal on surgical catheters and clothes;

[0013] S1103 excluding some liver regions affected by diseases on the CT value by setting a threshold slightly higher than the upper limit of the normal liver CT value. For example, the area density of HCC (hepatocellular carcinoma) is relatively large, the gray value is relatively low, and the gray value distribution is relatively dark;

[0014] S1104 performs preliminary extraction of bones after smoothing using Gaussian filtering and then performing Otsu threshold segmentation.

[0015] Further, the use of morphological operations and anatomical knowledge to denoise and completely segment bones described in step S120 specifically includes:

[0016] S1201 uses median filtering with a larger sampling window to eliminate isolated noise points;

[0017] S1202 processes the image in the X and Y directions using a Gaussian kernel to eliminate Gaussian noise.

[0018] S1203 uses median filtering with a smaller sampling window;

[0019] S1204 calculates the area of the connected region to remove objects with a smaller area;

[0020] S1205 uses morphological operations to delete objects that do not contain the structural element and delete narrow connections and small protrusions;

[0021] S1206 defines a two-dimensional elliptical region on each slice within the chest and abdominal cavities according to the inner contour of the peripheral fat region and the outer contour of the muscle region, and removes high-density calcified tissues within the abdominal cavity.

[0022] Further, the definition of a two-dimensional elliptical region described in step S1202 specifically includes:

[0023] (S1) Calculate the center point of the inner contour of the peripheral fat region or the outer contour of the muscle region, and the compression coefficient;

[0024] (S2) Calculate the center of the elliptical region by offsetting the center point obtained in step (S1) forward towards the body;

[0025] (S3) Define the major axis length according to the maximum distance between the left and right sides of the inner contour of the peripheral fat region;

[0026] (S4) Define the minor axis length according to the maximum distance between the front and back sides;

[0027] (S5) Remove the tissue within the ellipse in the CT images of the lower chest and upper abdomen.

[0028] Further, the use of statistical methods to extract the longitudinal boundary of liver segmentation based on bone information described in step S130 specifically includes:

[0029] S1301 analyzes the positional relationship between the bones and the liver according to anatomical knowledge;

[0030] S1302 calculates the bone pixel area, and the compactness coefficient cf = 1 - 4πa / P of the inner contour of the surrounding fat region2 ;

[0031] S1303. Extract the longitudinal boundary according to the change of the bone pixel area and the compactness coefficient.

[0032] Furthermore, the use of image processing technology to extract the transverse boundary of liver segmentation in step S140 specifically includes:

[0033] S1401. Combine each slice with its upper and lower slices to obtain more bone information;

[0034] S1402. Obtain the contour lines of each bone, and sort them after appropriate addition and deletion.

[0035] S1403. According to the sorted bone contour sequence, search for the distance between the nearest line segments of two adjacent contours to obtain the simple key points of the inner contour of the skeleton;

[0036] S1404. Fit the key points with a curve to draw the horizontal boundary. If there is no bone information, an estimated range is given.

[0037] Furthermore, the obtaining of the contour lines of each bone in step S1402, and sorting them after appropriate addition and deletion specifically includes:

[0038] (T1) Calculate the contour lines of each bone;

[0039] (T2) Use morphological operations to calculate the convex hull of the skeleton, and calculate the central coordinates of the minimum circumference of the convex hull;

[0040] (T3) Use the central coordinates obtained in step (T2) to divide the complete skeleton into four parts: upper left, lower left, upper right, and lower right;

[0041] (T4) According to the symmetric distribution characteristics of the bones, use the sternum as the base point to add bone information;

[0042] (T5) Sort the bones in a clockwise direction starting from the spine.

[0043] The present invention extracts bone information in CT images through image processing; uses morphological operations and anatomical knowledge to denoise and completely segment the bones; uses statistical methods to extract the longitudinal boundary of liver segmentation according to bone information; uses image processing technology to extract the transverse boundary of liver segmentation according to bone information.

[0044] The advantages of the present invention are: it helps to assist the liver segmentation algorithm, reduces the misrecognition rate of the liver area outside the abdominal wall, and reduces the processing time of the liver segmentation algorithm. Description of the Drawings

[0045] Figure 1It is a flowchart of the method of the present invention. Detailed implementation manners

[0046] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0047] As Figure 1 shown, in order to solve the above technical problems, Embodiment 1 of the present invention specifically provides a method for extracting the liver segmentation boundary based on bone information. The method includes: proposing a solution to the liver segmentation boundary problem of CT images.

[0048] Specifically, the method includes:

[0049] S110: Extract the bone information in the CT image through image processing.

[0050] Specifically, through image processing techniques, including but not limited to threshold segmentation, Gaussian filtering and other techniques, the bones are initially extracted.

[0051] S120: Use morphological operations and anatomical knowledge to denoise and completely segment the bones.

[0052] Specifically, image processing techniques with different sizes are used, including but not limited to median filtering, Gaussian filtering, etc. to process the bone image, calculate the area of the connected region to remove objects with smaller areas, use morphological operations to delete objects that do not contain structural elements, and delete narrow connections and small protrusions. According to the inner contour of the peripheral fat area and the outer contour of the muscle area, a two-dimensional elliptical region is defined on each slice in the chest and abdomen to remove high-density calcified tissues in the abdominal cavity.

[0053] S130: Use statistical methods to extract the longitudinal boundary of liver segmentation according to the bone information.

[0054] Specifically, analyze the positional relationship between the bones and the liver according to anatomical knowledge to obtain statistical laws, calculate the bone pixel area and the compactness coefficient cf = 1 - 4πa / P 2 of the inner contour of the surrounding fat area, and extract the longitudinal boundary according to the change of the bone pixel area and the compactness coefficient.

[0055] S140: Use image processing techniques to extract the transverse boundary of liver segmentation from the bone information.

[0056] Specifically, the upper and lower slices of each slice are combined to obtain more bone information. Appropriate operations are performed according to the contour lines of each bone, including but not limited to addition, deletion, and sorting. The distance between the nearest line segments of two adjacent contours is searched for in the processed bone contour sequence to obtain simple key points of the inner contour of the skeleton. The key points are curve-fitted to draw a horizontal boundary. If there is no bone information, an estimated range is given.

[0057] In summary, the method for extracting the liver segmentation boundary based on bone information provided by the present invention extracts bone information in a CT image through image processing; uses morphological operations and anatomical knowledge to denoise and completely segment the bones; uses statistical methods to extract the longitudinal boundary of liver segmentation based on bone information; and uses image processing techniques to extract the transverse boundary of liver segmentation based on bone information. This invention helps to assist the liver segmentation algorithm and can be used to reduce the misrecognition rate of the liver area outside the abdominal wall and reduce the processing time of the liver segmentation algorithm.

[0058] Inspired by the ideal embodiments of the present invention described above, through the above description, relevant staff can make various changes and modifications completely within the scope of the technical idea of this invention without deviation. The technical scope of this invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for extracting the liver segmentation boundary based on skeletal information, characterized in that It includes the following steps: S110, extracting bone information in the CT image through image processing; S120, denoising and segmenting bones by using morphological operations and anatomical knowledge; S130, extracting the longitudinal boundary of liver segmentation according to the bone information by using statistical methods; including: S1301, analyzing the positional relationship between the bone and the liver according to anatomical knowledge; S1302, calculating the bone pixel area and the compactness coefficient of the inner contour of the surrounding fat area; S1303, extracting the longitudinal boundary according to the change of the bone pixel area and the compactness coefficient; S140, extracting the transverse boundary of liver segmentation according to the bone information by using image processing technology; including: S1401, combining each slice with its upper and lower slices to obtain more bone information; S1402, obtaining the contour line of each bone, sorting after adding and deleting bone information; S1403, according to the sorted bone contour sequence, searching for the distance between the nearest line segments of two adjacent contours to obtain the simple key points of the inner contour of the skeleton; S1404, fitting the key points with a curve to draw the horizontal boundary, and giving an estimated range if there is no bone information.

2. The method for extracting the liver segmentation boundary based on skeletal information according to claim 1, wherein Step S110 includes: S1101, performing image preprocessing, mapping the CT value range from -1000 to 1000 to the pixel interval from 0 to 255; S1102, removing high-density noise by using a high threshold, including metals on surgical catheters and clothes; S1103, excluding the liver area affected by diseases on the CT value by setting a threshold higher than the upper limit of the normal liver CT value; S1104, using Gaussian filtering for smoothing and then performing Otsu threshold segmentation to initially extract bones.

3. The method for extracting the liver segmentation boundary based on skeletal information according to claim 1, wherein Step S120 includes: S1201, using median filtering with a large sampling window to eliminate isolated noise points; S1202, processing the image in the X and Y directions by using a Gaussian kernel to eliminate Gaussian noise; S1203, using median filtering with a small sampling window; S1204, calculating the area of the connected region to remove small-area objects; S1205, using morphological operations to delete objects without structural elements and deleting narrow connections and small protrusions; S1206, defining a two-dimensional elliptical region on each slice in the chest and abdominal cavities according to the inner contour of the peripheral fat area and the outer contour of the muscle area, and removing the high-density calcified tissue in the abdominal cavity.

4. The method for extracting the liver segmentation boundary based on bone information according to claim 3, wherein Step S1206 includes: (S1) calculating the center point and the compression coefficient of the inner contour of the peripheral fat area or the outer contour of the muscle area; (S2) calculating the center of the elliptical region by offsetting the center point obtained in step (S1) forward to the front of the body; (S3) defining the major axis length according to the maximum distance between the left and right sides of the inner contour of the peripheral fat area; (S4) defining the minor axis length according to the maximum distance between the front and back sides; (S5) removing the tissue inside the ellipse in the CT images of the lower chest and upper abdomen.

5. The method for extracting the liver segmentation boundary based on skeletal information according to claim 1, wherein Step S1402 includes: (T1) calculating the contour line of each bone; (T2) using morphological operations to calculate the convex hull of the skeleton and calculating the central coordinates of the minimum circumference of the convex hull; (T3) Divide the complete skeleton into four parts: upper left, lower left, upper right, and lower right using the central coordinates obtained in step (T2); (T4) Add bone information with the sternum as the base point according to the characteristic of symmetric bone distribution; (T5) Sort the bones in a clockwise direction starting from the spine.

Images