Flat-scanning CT coronary artery semi-automatic segmentation method and system based on region growth

Abstract

The invention relates to the technical field of medical image processing, in particular to a plain-scan CT coronary artery semi-automatic segmentation method and system based on region growth. The method comprises the following steps of: 1, performing data normalization according to a given window width and a given window level; step 2, contrast enhancement; step 3, performing median filtering; 4, seed point selection and boundary setting; step 5, carrying out region growth; step 6, filling the holes; and 7, editing by a physician. The boundary is automatically determined by fully utilizing the feature of density difference between the coronary artery of the plain-scan image and adjacent fat by means of region growth, and the technical bottleneck that a physician sketches the boundary layer by layer is broken through.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a method and system for semi-automatic segmentation of plain scan CT coronary arteries based on region growing. Background Art

[0002] Coronary heart disease is a major cardiovascular disease that endangers people's health, and its early screening and diagnosis are crucial. Coronary CT angiography (CCTA) is currently the preferred non-invasive examination method for clinical coronary heart disease, but due to risk factors such as adverse reactions to iodine contrast agents, it has not yet become a means of universal screening. It is of great significance to seek a convenient and safe screening method. Recent studies have shown that the CT plain scan pericoronary fat imaging genomics model has good efficacy in diagnosing non-calcified coronary plaques. However, coronary artery segmentation based on CT plain scan images is a technical bottleneck for the transformation of related research results. Existing literature all uses manual outlining to achieve plain scan CT coronary artery segmentation (i.e., manually outlining the coronary artery on CT plain scan images by comparing CCTA). This process is time-consuming and labor-intensive, and the physician's outlining burden is too heavy, which is not conducive to promotion and application. Therefore, a more efficient plain scan CT coronary artery segmentation method is urgently needed.

[0003] The region growing algorithm is a simple and effective segmentation method. By forming regions from sets of pixels with similar properties, it has been widely used in coronary segmentation based on CCTA images, but there has been no report on its application in coronary segmentation of plain CT images. Our experimental study found that the application of region growing in coronary segmentation of plain CT images is somewhat more special than CCTA, mainly because the density of coronary vessels is close to the surrounding soft tissue (myocardium), which makes direct application difficult. However, adding boundary settings and physician editing can better solve this technical problem. In this context, we propose a semi-automatic segmentation method for coronary arteries in plain CT based on region growing. This method is simple, universal, and physician-friendly. It is expected to help physicians segment coronary arteries more efficiently on plain CT images, and provide a basis for the further popularization and promotion of related clinical value research. Summary of the invention

[0004] In view of the above problems existing in the prior art, the present invention is proposed.

[0005] In order to solve the above technical problems, the present invention provides the following technical solution, a method for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing, comprising: performing data normalization according to a given window width and window level;

[0006] Perform contrast enhancement on the normalized image;

[0007] Median filtering is used to achieve image noise reduction;

[0008] Select seed points and set boundaries based on the denoised image, and perform region growing within the boundary setting range with the seed points as the starting point;

[0009] Fill all the holes in the region growing result;

[0010] The mask is edited by manual addition and deletion to finally obtain the segmentation result.

[0011] As a preferred solution of the method for semi-automatic segmentation of coronary arteries of plain scan CT based on region growing described in the present invention, wherein: the data normalization according to the given window width and window level includes reading the image data and header information of the DICOM file of the plain scan CT image, and converting the grayscale image of the original image into CT value;

[0012] The radiologist selects and records the window width and window level of the mediastinal window in the head information, and calculates the upper and lower limits based on the window width and window level;

[0013] For the image converted to CT value, each voxel value is mapped to the range between 0 and 1 from the lower limit to the upper limit.

[0014] As a preferred embodiment of the method for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing described in the present invention, the contrast enhancement is represented by further setting a certain interval range between 0 and 1, and mapping the normalized image to between 0 and 1 according to the interval range.

[0015] As a preferred solution of the method for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing described in the present invention, the seed point selection includes: a physician observes the course of the coronary artery to be segmented in the image after median filtering, sets a point inside the coronary artery to be segmented as a seed point, and records the position coordinates of the seed point;

[0016] The boundary setting includes the physician observing the course of the coronary artery to be segmented in the image after median filtering and setting a limited range for regional growth.

[0017] As a preferred solution of the semi-automatic segmentation method of coronary arteries of plain scan CT based on region growing described in the present invention, the region growing is performed within the set boundary range with the seed point as the starting point, including merging the voxels with grayscale similar to the seed voxel in the seed point and the surrounding neighborhood into the seed voxel area, using the new voxel as a new seed and continuing the iterative growth process until no more new seed voxels are generated, and finally the regional mask composed of all seed voxels is the regional growing result.

[0018] As a preferred solution of the method for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing described in the present invention, wherein: all the hollow regions existing in the filling region growth result include assigning the values ​​of all the hollow regions to 1.

[0019] As a preferred embodiment of the semi-automatic segmentation method of coronary arteries of plain scan CT based on region growing described in the present invention, the final segmentation result is expressed as follows: the final segmentation result includes that the value of the area that the physician determines needs to be added after observation is assigned to 1, the value of the area that the physician determines needs to be deleted after observation is assigned to 0, and the value of the area that does not belong to the area that the physician determines needs to be added after observation, nor the area that the physician determines needs to be deleted after observation, is assigned the result after filling the hole.

[0020] As a preferred solution of the semi-automatic segmentation system of plain scan CT coronary arteries based on region growing described in the present invention, it includes: an image extraction module, a data normalization module, a contrast enhancement module, a median filtering module, a seed point selection and boundary setting module, a region growing module, a cavity filling module, and a physician editing module;

[0021] The image extraction module extracts the original image and determines the window width and window position;

[0022] The data normalization module performs data normalization according to a given window width and window level;

[0023] The contrast enhancement module performs a linear transformation on the normalized image to enhance the contrast;

[0024] The median filtering module performs median filtering on the contrast-enhanced image to achieve image noise reduction;

[0025] The seed point selection and boundary setting module is used by the physician to observe the course of the coronary artery to be segmented in the median filtered image and set the seed point and the limited range of the region growth;

[0026] The region growing module performs region growing on the median filtered image within a set boundary range with the seed point as the starting point to obtain a region growing result;

[0027] The hole filling module fills all the hole regions existing in the region growing result;

[0028] In the physician editing module, the physician manually adds, deletes and edits the region growing result after filling the holes to obtain the final segmentation result.

[0029] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein when the processor executes the computer program, the processor implements the steps of any one of the methods for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing.

[0030] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of any one of the methods for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing are implemented.

[0031] The beneficial effects of the present invention are as follows: by using regional growth, the characteristics of the difference in density between the coronary artery and the adjacent fat in the plain scan image are fully utilized to automatically determine the boundary, thus breaking through the technical bottleneck of physicians outlining the boundary layer by layer. Compared with the layer-by-layer outlining method ("addition") used in actual clinical research work, the present invention provides a new idea, that is, the physician only needs to perform an editing process with deleting the region as the main operation ("subtraction") based on the preliminary segmentation results obtained by regional growth, which can greatly reduce the workload and time and improve the physician's research work experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0033] Figure 1 A schematic flow chart of a method for semi-automatic segmentation of coronary arteries in plain scan CT based on region growing provided in one embodiment of the present invention.

[0034] Figure 2 The image of the coronary artery calcium score CT examination image of a patient adopted in the embodiment of the present invention after normalization processing (i.e., the output image of step 1), showing a total of 4 cross-sectional planes (4 columns);

[0035] Figure 3 This is the contrast-enhanced image obtained through step 2 in the embodiment of the present invention.

[0036] Figure 4 This is the median filtered image obtained in step 3 in the embodiment of the present invention.

[0037] Figure 5 This is the region growing result (ie, mask1) obtained through step 5 in the embodiment of the present invention.

[0038] Figure 6 This is the result of filling the holes (ie, mask2) obtained through step 6 in the embodiment of the present invention.

[0039] Figure 7 The result after doctor editing (ie, mask 3) obtained in step 7 of the embodiment of the present invention is the final segmentation result. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0041] Example 1

[0042] Reference Figure 1-Figure 7 , which is the first embodiment of the present invention, provides a semi-automatic segmentation method of coronary arteries in plain scan CT based on region growing, such as Figure 1 As shown, including:

[0043] A semi-automatic segmentation method for coronary arteries in plain scan CT based on region growing, such as Figure 1 As shown, the steps include: step 1, normalizing data according to a given window width and window position; step 2, contrast enhancement; step 3, median filtering; step 4, seed point selection and boundary setting; step 5, region growing; step 6, filling holes; and step 7, physician editing.

[0044] The specific process is as follows:

[0045] Step 1: normalize the data according to the given window width and window level; the specific process is:

[0046] ① Image information reading and processing, that is, reading the original image and header information, and converting the original image grayscale value into CT value;

[0047] Specifically, input the DICOM image data of a patient's plain scan coronary calcification integral CT examination. Read the original image, and assume that the read original image is I. I is a three-dimensional matrix of 512×512×256, and 512, 512, and 256 represent the number of elements of I in the X, Y, and Z directions, respectively. Read the window width and window position information in the image file header information. In this embodiment, the window width and window position are 800 and 100, respectively. The radiologist confirms that there is only one pair of window width and window position information, and it is a mediastinal window, so the window width (WW) is recorded as 800 and the window position (WC) is recorded as 100. Read the two parameters of rescale slope and rescale intercept in the header information. In this embodiment, rescale slope=1, rescale intercept=-1024. Convert the grayscale value of the original image to CT value, and assume that the image after conversion to CT value is I. 2 , I 2 It is a three-dimensional matrix of 512×512×256. According to the calculation formula:

[0048] I2 =I×rescale slope+rescale intercept

[0049] In this embodiment, I 2 The calculation method can be expressed as:

[0050] I 2 =I×1+(-1024)

[0051] ② Calculate the upper and lower bounds. The algorithm is: upper bound = window position + 0.5 × window width, lower bound = window position - 0.5 × window width;

[0052] Specifically, let the upper bound be p and the lower bound be q, according to the calculation formula:

[0053] p=WC+0.5×WW, q=WC-0.5×WW

[0054] In this embodiment, the upper bound (p) = 100 + 0.5 × 800 = 500,

[0055] Lower bound (q) = 100 - 0.5 × 800 = -300.

[0056] ③ Data normalization, that is, for the image converted into CT value, each voxel value is mapped to the interval range from the lower limit to the upper limit to between 0 and 1 (that is, the normalized image). The algorithm is: the grayscale value greater than the upper limit is mapped to 1, and the grayscale value less than the lower limit is mapped to 0. The grayscale values ​​between the lower limit and the upper limit are linearly mapped to between 0 and 1.

[0057] Specifically, let the normalized image be I 3 , I 3 is a three-dimensional matrix of 512×512×256. In this embodiment, the upper bound p=500 and the lower bound q=-300, so I 3 The calculation method is: traverse the matrix I 2 For each element (X=x,Y=y,Z=z), for elements greater than 500, I 3 The value of the corresponding position is assigned to 1; for elements less than -300, I 3 The value of the corresponding position is assigned to 0; for elements in the range [-300,500], I 3 The value of the corresponding position is assigned to I 2 The value is linearly mapped from [-300,500] to [0,1]. The calculation process can be expressed as:

[0058]

[0059] Among them, I 2 (x,y,z) represents the matrix I 2The numerical value of the element at position X = x, Y = y, Z = z in I 3 (x, y, z) represents the matrix I 3 The numerical value of the element at position X = x, Y = y, Z = z in it.

[0060] After the above processing of this embodiment, the normalized image (i.e., I 3 ) is as Figure 2 shown.

[0061] For the "plain CT coronary artery segmentation" scenario, performing this step helps map the image grayscale to between 0 and 1, facilitating the subsequent steps. In this invention, this step normalizes the data according to the mediastinal window selected by the radiologist, making the grayscale distribution of the normalized image more in line with the actual clinical perspective, which is a feature of this invention.

[0062] Step 2, contrast enhancement; the specific process is as follows:

[0063] To enhance the image contrast, a certain range within 0 to 1 is further set, and the normalized image is mapped to between 0 and 1 according to this range (i.e., the contrast-enhanced image). The algorithm is: the grayscale value greater than the upper bound of this range is mapped to 1, the grayscale value less than the lower bound of this range is mapped to 0, and the grayscale values between the lower bound and the upper bound of this range are linearly mapped to between 0 and 1.

[0064] Specifically, let the contrast-enhanced image be I 4 , and I 4 is a three-dimensional matrix of 512×512×256. Let a certain range within 0 to 1 be further set as [v, u], where 0 < v < u < 1. In this embodiment, v = 0.2 and u = 0.8 are specified. The calculation method of I 4 is: traverse each element (X = x, Y = y, Z = z) of the matrix I 3 . For the element greater than 0.8, the value at the corresponding position of I 4 is assigned 1; for the element less than 0.2, the value at the corresponding position of I 4 is assigned 0; for the elements in the range [0.2, 0.8], the value at the corresponding position of I 4 is assigned the value of I 3 linearly mapped from [0.2, 0.8] to [0, 1]. The calculation process can be expressed as:

[0065]

[0066] where, I 3 (x, y, z) represents the numerical value of the element at position X = x, Y = y, Z = z in the matrix I 3 , and I 4(x,y,z) represents the matrix I 4 The value of the element at position X=x, Y=y, Z=z.

[0067] After the above processing, the contrast-enhanced image (i.e., I 4 )like Figure 3 shown.

[0068] For the "plain scan CT coronary segmentation" scenario, since the density of coronary vessels on plain scan CT images is close to that of the surrounding soft tissue (myocardium), the contrast between the coronary artery and the adjacent fat tissue is much smaller than that on CTA images, which is the particularity of "plain scan CT coronary segmentation" compared to "CTA coronary segmentation". The present invention performs this step to help improve the contrast between the coronary artery and the adjacent fat tissue, which may improve the accuracy of the subsequent regional growth to determine the boundary.

[0069] Step 3: median filtering. The specific process is as follows:

[0070] The traditional median filter is used to achieve image denoising. Median filtering is a nonlinear smoothing technique that can eliminate isolated noise points and effectively remove image noise while retaining edge information. Its basic principle is to take out the pixel values ​​of all pixels in the neighborhood window of a pixel point and sort them, and take the middle value as the value of the pixel point. For three-dimensional images, the specific method is to use a three-dimensional sliding template to rearrange the voxel grayscale values ​​in the template in ascending (or descending) order to generate a one-dimensional data sequence, and take the median value instead of the original voxel grayscale value.

[0071] Specifically, let the image after median filtering be I 5 , I 5 is a 3D matrix of 512×512×256. Let the 3D sliding template be W, and W is a 3×3×3 (voxel 3 ) cube. 5 The calculation method is: traverse the matrix I 4 For each element (X=x, Y=y, Z=z) except the boundary points, the original value of the element is replaced by the median of the 3×3×3 neighborhood values ​​of the element. The calculation process can be expressed as:

[0072] I 5 (x,y,z)=med{I 4 (x+ξ,y+η,z+ζ) , (ξ,η,ζ∈W)}

[0073] Among them, I 4 (x+ξ,y+η,z+ζ) represents the matrix I 4 The element value at position X=x+ξ, Y=y+η, Z=z+ζ, ζ, η, ζ are all integers in the range [-1, 1], and W represents the size of 3×3×3 (voxel3 ) three-dimensional sliding template, med represents the median operation, I 5 (x,y,z) represents the matrix I 5 The value of the element at position X=x, Y=y, Z=z.

[0074] After the above processing, the median filtered image (i.e., I 5 )like Figure 4 shown.

[0075] For the "plain scan CT coronary segmentation" scenario, since plain scan CT images (such as calcification integration) usually have relatively large noise, the present invention can reduce image noise by performing this step, thereby reducing the potential impact of image noise on subsequent segmentation.

[0076] Step 4: Seed point selection and boundary setting, specifically:

[0077] ① Seed point selection: the physician observes the course of the coronary artery to be segmented in the median filtered image, sets a point inside the coronary artery to be segmented as the seed point, and records the position coordinates of the seed point;

[0078] Specifically, let the seed point determined by the doctor be P 0 , then P 0 The position coordinates can be expressed as: (x 0 ,y 0 ,z 0 ). In this embodiment, the coronary artery to be segmented is "the left coronary artery starts from the left main trunk, and starts from 10 mm at the coronary artery opening to the left anterior descending branch for a continuous 40 mm". After observing the course of the coronary artery to be segmented in the image, the doctor determines the seed point P 0 The position coordinates (x 0 ,y 0 ,z 0 ) is: x 0 =240,y 0 =227,z 0 =47.

[0079] ② Boundary setting: the physician observes the course of the coronary artery to be segmented in the median filtered image and sets the limited range of the region growing, i.e., a cube containing the region of interest of the coronary artery to be segmented. The cube can be determined by the two vertices at both ends of its diagonal line, and the position coordinates of the two vertices are recorded;

[0080] Specifically, let the two vertices at both ends of the diagonal of the limited range (cube) determined by the doctor be P 1 and P 2 , then P 1 and P 2 The position coordinates can be expressed as: (x 1 ,y 1,z 1 ) and (x 2 ,y 2 ,z 2 ). In this embodiment, after observing the image, the physician determines the vertex P 1 The position coordinates (x 1 ,y 1 ,z 1 ) is: x 1 =211,y 1 =178,z 1 =40, vertex P 2 The position coordinates (x 2 ,y 2 ,z 2 ) is: x 2 =274,y 2 =262,z 2 =60.

[0081] For the "plain scan CT coronary segmentation" scenario, since the density of coronary vessels on plain scan CT images is close to that of the surrounding soft tissue (myocardium), if the boundary is not set, the regional growth result will contain a large number of non-coronary areas, and the iteration cannot be completed quickly. This is the particularity of "plain scan CT coronary segmentation" compared to "CTA coronary segmentation". The boundary setting added by the present invention limits the scope of regional growth and avoids unnecessary growth time consumption, which is a difference between the present invention and other inventions.

[0082] Step 5: Region growing. The specific process is as follows:

[0083] Starting from the seed point, region growing is performed within the boundary setting range. Specifically, the voxels with similar grayscale to the seed voxel in the seed point and the surrounding neighborhood are merged into the seed voxel region, and these new voxels are used as new seeds to continue the above process until no more new seed voxels are generated. Finally, the regional mask composed of all seed voxels is the result of region growing. [Note: Mask is an important concept in the field of image processing. It is used to specify a region of an image. It is usually a binary image of the same size as the original image, in which the selected area is marked as 1 and the rest of the area is marked as 0.]

[0084] Specifically, let the region growing result be M 1 , M 1 It is a three-dimensional matrix of 512×512×256. 1 The calculation method is:

[0085] ① Define matrix J to store the seed voxel region. J is a 3D matrix of 512×512×256. Initially, J contains only the seed point P determined by the physician. 0 The value of the position is 1, and the rest are 0. It can be expressed as:

[0086]

[0087] Among them, J(x,y,z) represents the value of the element at position X=x,Y=y,Z=z in the matrix J.

[0088] ②Define variable K to store the seed voxel region I 5 The mean value changes as the number of seed voxels increases. It can be expressed as:

[0089] R = {(x,y,z)|J(x,y,z) = 1}

[0090]

[0091] Among them, J(x,y,z) represents the value of the element at position X=x,Y=y,Z=z in matrix J, and R is The area that satisfies J(x,y,z)=1, I 5 (x,y,z) represents the matrix I 5 The value of the element at position X=x, Y=y, Z=z in the set. card represents the operation of finding the number of elements in the set.

[0092] ③ For an element whose matrix J value is 1, if the element is located within the aforementioned limited range (cube), search the 3×3×3 neighborhood of the element. If an element position in the neighborhood corresponds to a J value of 0 and corresponds to the matrix I 5 Value and seed voxel region I 5 If the absolute value of the mean (K) difference is less than the threshold α (the threshold α is set to 0.1 in this embodiment), the J value of the element position is set to 1, and the grayscale mean K of the seed voxel region is updated. It can be expressed as:

[0093] J(x+ξ,y+η,z+ζ)=1, when both satisfy

[0094] Among them, J(x,y,z) represents the value of the element at position X=x,Y=y,Z=z in the matrix J, ξ,η,ζ are all integers in the range [-1,1], and W represents the size of 3×3×3 (voxels). 3 ), J(x+ξ,y+η,z+ζ) represents the value of the element at position X=x+ξ,Y=y+η,Z=z+ζ in matrix J, I 5 (x+ξ,y+η,z+ζ) represents the matrix I 5 The value of the element whose position is X=x+ξ,Y=y+η,Z=z+ζ.

[0095] ④Continuously repeat the previous step, i.e. step ③, until no more new seed voxels are generated. Finally, the regional mask composed of all seed voxels is the result of regional growth (obtaining mask1). It can be expressed as:

[0096] M 1 =J, when count = 0

[0097] Here, count represents the number of new seed voxels generated compared to the previous iteration.

[0098] After the above processing, the region growing result (i.e., mask1) obtained in this embodiment is as follows: Figure 5 shown.

[0099] For the "plain scan CT coronary artery segmentation" scenario, the implementation of this step is aimed at obtaining preliminary segmentation results.

[0100] Step 6: Fill the holes. The specific process is as follows:

[0101] Fill all the empty areas in the region growing result, that is, assign the values ​​of all the empty areas to 1 (obtain mask2).

[0102] Specifically, suppose the result after filling the hole is M 2 , M 2 is a three-dimensional matrix of 512×512×256; let the set of holes in the region growing result be S H . M 2 The calculation method is:

[0103]

[0104] Among them, M 1 (x,y,z) represents the matrix M 1 The value of the element at position X=x, Y=y, Z=z, M 2 (x,y,z) represents the matrix M 2 The value of the element at position X=x, Y=y, Z=z.

[0105] After the above processing, the region growth result after filling the holes (i.e., mask2) is obtained as follows: Figure 6 shown.

[0106] For the "plain scan CT coronary artery segmentation" scenario, this step helps to make the segmentation results more complete and reduces the workload of subsequent physician editing to a certain extent.

[0107] Step 7, physician editing, specifically:

[0108] The physician observes the course of the coronary artery to be segmented in the image, overlays the result of the previous step (mask2) with the original image for observation, and edits the mask in a manual addition and deletion manner to finally obtain the segmentation result (obtaining mask3).

[0109] Specifically, let the result after the physician's editing be M 3 , M 3 is a three-dimensional matrix of 512×512×256; let the area (set of coordinate points) that the physician deems needs to be added after observation be S A , and the area (set of coordinate points) that is considered needs to be deleted be S D . The calculation method of M 3 is as follows:

[0110]

[0111] where M 2 (x,y,z) represents the element value at the position of X = x, Y = y, Z = z in the matrix M 2 , and M 3 (x,y,z) represents the element value at the position of X = x, Y = y, Z = z in the matrix M 3 .

[0112] In this embodiment, the physician uses the ROI drawing tool (ITK-SNAP) to overlay mask2 with the original image for observation, and edits the mask in a manual addition and deletion manner. The obtained mask3, which is the final segmentation result, is as Figure 7 shown.

[0113] For the "plain CT coronary artery segmentation" scenario, since the density of coronary artery vessels on plain CT images is close to that of surrounding soft tissues (myocardium), the preliminary segmentation result obtained by region growing will inevitably contain non-coronary artery structures. Therefore, it is necessary to combine with the physician for quality control, which is the particularity of "plain CT coronary artery segmentation" compared with "CTA coronary artery segmentation". This step solves the technical problem of the difficulty in directly applying region growing in the "plain CT coronary artery segmentation" scenario by the physician editing the segmentation result, which is a difference of the present invention compared with other inventions.

[0114] The advantages of the present invention over the prior art are: by using regional growth, the characteristics of the difference in density between the coronary artery and the adjacent fat in the plain scan image are fully utilized to automatically determine the boundary, breaking through the technical bottleneck of the physician outlining the boundary layer by layer. In addition, the specific algorithm of the present invention also has the following advantages: ① Through contrast enhancement, the contrast between the coronary artery and the adjacent fat tissue is improved, and the boundary determination after execution may be more accurate; ② Through median filtering, image noise and its potential impact on segmentation are reduced; ③ By filling the holes, the coronary segmentation is made more complete, which reduces the workload of subsequent physician editing to a certain extent; ④ Through boundary setting and physician editing, the technical problem that the density of coronary vessels in plain scan CT is close to the surrounding soft tissue (myocardium) and the direct application of regional growth is difficult is solved.

[0115] Example 2

[0116] The second embodiment of the present invention is different from the previous embodiment in that:

[0117] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0118] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0119] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0120] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0121] Example 3

[0122] The third embodiment of the present invention provides a semi-automatic segmentation system for coronary arteries of plain scan CT based on region growing, which is characterized by comprising an image extraction module, a data normalization module, a contrast enhancement module, a median filtering module, a seed point selection and boundary setting module, a region growing module, a cavity filling module, and a physician editing module;

[0123] The image extraction module extracts the original image and determines the window width and window position;

[0124] The data normalization module performs data normalization according to a given window width and window level;

[0125] The contrast enhancement module performs a linear transformation on the normalized image to enhance the contrast;

[0126] The median filtering module performs median filtering on the contrast-enhanced image to achieve image noise reduction;

[0127] The seed point selection and boundary setting module is used by the physician to observe the course of the coronary artery to be segmented in the median filtered image and set the seed point and the limited range of the region growth;

[0128] The region growing module performs region growing on the median filtered image within a set boundary range with the seed point as the starting point to obtain a region growing result;

[0129] The hole filling module fills all the hole regions existing in the region growing result;

[0130] In the physician editing module, the physician manually adds, deletes and edits the region growing result after filling the holes to obtain the final segmentation result.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A semi-automatic coronary artery segmentation method based on region growing in plain scan CT, characterized by: include, Normalize the data according to the given window width and window level; Perform contrast enhancement on the normalized image; Median filtering is used to achieve image noise reduction; Select seed points and set boundaries based on the denoised image, and perform region growing within the boundary setting range with the seed points as the starting point; Fill all the holes in the region growing result; The mask is edited by manual addition and deletion to finally obtain the segmentation result.

2. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 1, characterized in that: The data normalization according to the given window width and window level includes reading the image data and header information of the DICOM file of the plain scan CT image, and converting the grayscale image of the original image into a CT value; The radiologist selects and records the window width and window level of the mediastinal window in the head information, and calculates the upper and lower limits based on the window width and window level; For the image converted to CT value, each voxel value is mapped to the range between 0 and 1 from the lower limit to the upper limit.

3. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 2, characterized in that: The contrast enhancement is expressed as setting an interval range between 0 and 1, and mapping the normalized image to between 0 and 1 according to the interval range.

4. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 3, characterized in that: The seed point selection includes: the physician observes the course of the coronary artery to be segmented in the median filtered image, sets a point inside the coronary artery to be segmented as a seed point, and records the position coordinates of the seed point; The boundary setting includes the physician observing the course of the coronary artery to be segmented in the image after median filtering and setting a limited range for regional growth.

5. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 4, characterized in that: The method of performing region growing within the set boundary range with the seed point as the starting point includes merging voxels with grayscales similar to the seed voxel in the seed point and the surrounding neighborhood into the seed voxel region, using the new voxel as a new seed and continuing the iterative growth process until no more new seed voxels are generated, and finally the regional mask composed of all seed voxels is the regional growing result.

6. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 5, characterized in that: All the hole regions existing in the filling region growth result include assigning values ​​of all the hole regions to 1.

7. The method for semi-automatic coronary artery segmentation based on region growing in plain scan CT according to claim 6, characterized in that: The final segmentation result includes that the value of the area that the doctor determines needs to be added after observation is assigned to 1, the value of the area that the doctor determines needs to be deleted after observation is assigned to 0, and the value of the area that does not belong to the area that the doctor determines needs to be added after observation, and the area that does not belong to the area that the doctor determines needs to be deleted after observation, is assigned the result after filling the hole.

8. A system based on the region growing-based semi-automatic segmentation method for coronary arteries in plain scan CT according to any one of claims 1 to 7, characterized in that: Including image extraction module, data normalization module, contrast enhancement module, median filtering module, seed point selection and boundary setting module, region growing module, hole filling module, and physician editing module; The image extraction module extracts the original image and determines the window width and window position; The data normalization module performs data normalization according to a given window width and window level; The contrast enhancement module performs a linear transformation on the normalized image to enhance the contrast; The median filtering module performs median filtering on the contrast-enhanced image to achieve image noise reduction; The seed point selection and boundary setting module is used by the physician to observe the course of the coronary artery to be segmented in the median filtered image and set the seed point and the limited range of the region growth; The region growing module performs region growing on the median filtered image within a set boundary range with the seed point as the starting point to obtain a region growing result; The hole filling module fills all the hole regions existing in the region growing result; In the physician editing module, the physician manually adds, deletes and edits the region growing result after filling the holes to obtain the final segmentation result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images