A method for ascending aorta segmentation based on coronary CTA images

Abstract

This invention provides a method for ascending aorta segmentation based on CTA images. First, the CTA image data is preprocessed to determine the region of interest (ROI) of the ascending aorta, reducing computational load and improving efficiency. Then, using Hough circle detection and pixel features of the ascending aorta, the position coordinates and radius of the ascending aorta are determined, resulting in an initial ascending aorta region. Next, the initial ascending aorta region is subjected to erosion and dilation operations of varying intensities, serving as foreground and background masks for a graph cutting method. An optimized graph cutting method is then used for segmentation, retaining the segmentation result corresponding to the shortest path. This invention exhibits better segmentation capabilities for blurred boundaries of the target; compared to traditional region growing methods, this invention is faster and requires no manual interaction; compared to deep learning-based segmentation methods, this invention does not require a large amount of data support and is less costly.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method for ascending aorta segmentation based on coronary CTA images. Background Technology

[0002] Cardiovascular disease, primarily caused by insufficient blood supply to the myocardium leading to impaired heart rhythms, has become the leading cause of death and is a major concern for human health. Extensive clinical evidence indicates that cardiovascular disease is primarily caused by coronary artery abnormalities. The coronary arteries are blood vessels that surround the heart, originating from the root of the ascending aorta and branching into left and right branches.

[0003] Coronary CT angiography (CTA) is increasingly used to diagnose cardiovascular diseases. The coronary arteries originate from the root of the ascending aorta and divide into left and right branches. In cardiovascular analysis and diagnosis, the ascending aorta plays a crucial role in the localization and segmentation of the coronary arteries. Fully automated ascending aortic segmentation is a prerequisite for automated coronary artery segmentation. By using the location information of the ascending aorta, the locations of the openings of the left and right coronary arteries can be identified, and the vessels can be named, facilitating better analysis of cardiovascular tissue by clinicians. Simultaneously, segmenting the ascending aorta can reduce the difficulty of segmenting the left and right coronary arteries, enabling fully automated coronary artery segmentation and providing meaningful diagnostic and treatment information for the clinical diagnosis of coronary arteries.

[0004] Currently, ascending aorta segmentation methods mainly include region-growing-based methods and deep learning-based methods. The shortcomings of region-growing-based methods are: ① manual selection of seed points, adding an extra manual interaction process, preventing full automation; ② easy leakage into other tissues during segmentation, causing adhesion with other organs; ③ prone to incomplete segmentation, significantly reducing segmentation accuracy. Deep learning-based segmentation methods require large amounts of data, are costly, and have certain limitations in practical applications. Summary of the Invention

[0005] To overcome the above-mentioned technical defects, the purpose of this invention is to provide a method for ascending aorta segmentation based on coronary CTA images that has high computational efficiency and can effectively segment blurred boundaries.

[0006] This invention discloses a method for ascending aorta segmentation based on coronary CTA images, comprising: determining a region of interest (ROI) of the ascending aorta in the coronary CTA image based on the lung region; determining an initial location region of the ascending aorta using the Hough circle detection method, wherein the initial location region includes center coordinates and radius; performing erosion and dilation operations of different intensities on the initial location region of the ascending aorta as foreground and background masks for a graph cutting method; segmenting the initial location region of the ascending aorta using the graph cutting method, and retaining the segmentation result corresponding to the shortest path as the segmentation result of the ascending aorta.

[0007] Preferably, determining the region of interest (ROI) of the ascending aorta in the coronary CTA image based on the lung region includes: extracting the lung region from the coronary CTA image; calculating the centroid of the lung region based on CT values; and defining the region within a preset range around the centroid as the ROI of the ascending aorta.

[0008] Preferably, determining the initial location region of the ascending aorta using the Hough circle detection method includes: calculating n first-class detection circles that may contain the region of the ascending aorta using the Hough circle detection method, and determining the center coordinates and radius of the detection circles; n≥3; selecting the second-class detection circle that is most likely to be the ascending aorta among the first-class detection circles based on the characteristic parameters of the ascending aorta; and using the maximum connected component label, selecting the center coordinates and radius of the detection circle with the largest area among the second-class detection circles as the starting position center coordinates and starting position radius of the ascending aorta.

[0009] Preferably, the step of selecting the second type of detection circle most likely to be the ascending aorta from the first type of detection circles using the characteristic parameters of the ascending aorta includes: selecting detection circles in the first type of detection circles whose Hough cumulative weight is greater than or equal to a preset weight value, selecting detection circles in the first type of detection circles whose average pixel gray value is greater than or equal to a preset pixel gray value, and selecting detection circles in the first type of detection circles whose pixel gray standard deviation is greater than or equal to a preset standard deviation, thereby selecting the second type of detection circle most likely to be the ascending aorta from the first type of detection circles.

[0010] Preferably, after selecting the center coordinates and radius of the detection circle with the largest area among the second type of detection circles as the starting position center coordinates and starting position radius of the ascending aorta, the method further includes: starting from the starting position center coordinates of the ascending aorta, calculating the gradient energy of each cross section of the ascending aorta; within a first preset neighborhood of the starting position center coordinates of the ascending aorta, and within a second preset neighborhood of the starting position radius, taking the coordinates corresponding to the maximum gradient energy as the center position coordinates of the cross section of the ascending aorta, and the radius corresponding to the maximum gradient energy as the radius of each cross section of the ascending aorta.

[0011] Preferably, calculating the energy of each cross-section of the ascending aorta includes: the gradient energy corresponding to the i-th pixel. Among them, GE xi Let be the gradient value of the i-th pixel in the x-direction; GE yi Let be the gradient value of the i-th pixel in the y-direction; GE zi Let be the gradient value of the i-th pixel in the z-direction.

[0012] Preferably, the step of performing erosion and dilation operations of different intensities on the initial location region of the ascending aorta as foreground and background masks for the graph tiling method; segmenting the initial location region of the ascending aorta using the graph tiling method and retaining the segmentation result corresponding to the shortest path as the segmentation result of the ascending aorta includes: determining the initial ascending aorta mask based on the position coordinates and radius of the ascending aorta; eroding the initial ascending aorta mask as the foreground mask; performing dilation operations of different intensities on the initial ascending aorta mask as the background mask; segmenting the initial location region of the ascending aorta using the graph tiling method; and calculating the weight energy CapSourse[x] of the foreground mask. i y i , z i ] and the weight energy CapSink[x] of the background mask i y i , z i ]; Calculate energy E using an energy function; Select the minimum energy value min(E[x]); i y i , z i The segmentation result corresponding to the above is obtained by continuously optimizing the foreground and background masks and using the graph cutting method to segment the initial position region of the ascending aorta until the rate of change of the minimum energy value between two adjacent segments is less than 0.1. The segmentation result is then smoothed at the edges to obtain the segmentation result of the ascending aorta.

[0013] Preferably, the calculation of the weight energy CapSourse[x] of the foreground mask... i y i , z i ] and the weight energy CapSink[x] of the background mask i y i , z i ]include:

[0014] Calculate the gray-level histogram of the image in the foreground mask, and count the frequency p of each pixel gray-level value in the foreground mask;

[0015] If ForeMask[x] i yi , z i If ] = 1, then CapSourse[x] i y i , z i ] = MaxValue;

[0016] If BackMask[x i y i , z i If ] = 1, then CapSourse[x] i y i , z i ] = 0;

[0017] Otherwise, CapSourse[x i y i , z i ] = p[ImageDate[x i y i , z i [ImageDate[x] i y i , z i [] represents the grayscale value of the i-th pixel.

[0018] If BackMask[x i y i , z i If ] = 1, then CapSink[x i y i , z i ] = MaxValue;

[0019] Otherwise, CapSink[x i y i , z i ] = 0.

[0020] Preferably, the calculation of the weight energy CapSourse[x] of the foreground mask... i y i , z i ] and the weight energy CapSink[x] of the background mask i y i , z i It also includes calculating the weighted energy of the 6-neighborhood of all pixels:

[0021] The weight energy CapLee[x] of one neighborhood i y i , z i ]for:

[0022] If ForeMask[x]i y i , z i ] = 1 & ForeMask[x i -1, y i , z i ] = 1, or BackMask[x i y i , z i ] = 1 & BackMask[x i -1, y i , z i If ] = 1, then CapLee[x] i y i , z i ] = MaxValue;

[0023] otherwise,

[0024] Where MaxValue is the maximum value, (x i y i , z i ) represents the position coordinates of the i-th pixel, [ImageDate[x i y i , z i ]] represents the grayscale value of the i-th pixel.

[0025] Preferably, the calculation of energy E using the energy function includes:

[0026] Energy E[x i y i , z i ]=λR[x i y i , z i ]+B[x i y i , z i ];

[0027] Wherein, R[x i y i , z i B[x] represents the sum of the weighted energies of the foreground and background masks. i y i , z i ] represents the total weight energy among 6 neighboring pixels, and λ is the influence factor of the weight energy of the region and the weight energy between adjacent pixels.

[0028] Compared with existing technologies, the above technical solution has the following advantages:

[0029] 1. This invention first preprocesses the CTA image data to determine the region of interest of the ascending aorta, thereby reducing computational load and improving efficiency. Then, using Hough circle detection and pixel features of the ascending aorta, the position coordinates and radius of the ascending aorta are determined to obtain the initial ascending aorta region. Next, the initial ascending aorta region is subjected to erosion and dilation operations of different intensities as foreground and background masks for the graph cutting method. The graph cutting method is then used for segmentation, retaining the segmentation result corresponding to the shortest path.

[0030] 2. Compared with traditional graph cutting methods, this invention has better segmentation capabilities for fuzzy target boundaries; compared with traditional region growing methods, this invention is faster and does not require manual interaction; compared with deep learning-based segmentation methods, this invention does not require a large amount of data support and is less expensive. Attached Figure Description

[0031] Figure 1 A flowchart of the ascending aorta segmentation method based on coronary CTA images provided by the present invention;

[0032] Figure 2 A flowchart for obtaining the region of interest provided by this invention;

[0033] Figure 3 A flowchart illustrating the determination of the initial location region of the ascending aorta using the Hough circle detection method provided by the present invention;

[0034] Figure 4 A flowchart of segmentation using the graphical cutting method provided by the present invention;

[0035] Figure 5 shows the region of interest of the ascending aorta defined in a coronary CTA image provided by the present invention, where (a) is a coronary CTA image and (b) is the defined region of interest.

[0036] Figure 6 shows the detection circles determined by the Hough circle detection method provided by the present invention, where (a) is the first type of detection circle and (b) is the second type of detection circle;

[0037] Figure 7 shows an example of a coronary CTA image;

[0038] Figure 8 shows the segmentation results of the prior art as an example;

[0039] Figure 9 shows an example of the segmentation results of the present invention;

[0040] Figure 10 shows the results of ascending aortic segmentation in different sections provided by the present invention, where (a) is a transverse section, (b) is a sagittal section, and (c) is a coronal section. Detailed Implementation

[0041] The advantages of the present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments.

[0042] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0043] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0044] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0045] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0046] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0047] In the following description, suffixes such as "module," "part," or "unit" used to denote elements are used only for the convenience of the description of the invention and have no specific meaning in themselves. Therefore, "module" and "part" can be used interchangeably.

[0048] See appendix Figure 1 This invention discloses a method for ascending aortic segmentation based on coronary CTA images, comprising:

[0049] S100. Determine the region of interest of the ascending aorta in coronary CTA images based on the lung region;

[0050] S200. The initial location region of the ascending aorta is determined using the Hough circle detection method. The initial location region includes the center position coordinates and radius.

[0051] S300. Perform erosion and dilation operations of different strengths on the initial location region of the ascending aorta, which serve as the foreground and background masks for the graph cutting method; use the graph cutting method to segment the initial location region of the ascending aorta, and retain the segmentation result corresponding to the shortest path as the segmentation result of the ascending aorta.

[0052] First, in the entire coronary CTA image, the ascending aorta region occupies only a small portion and is surrounded by the lung region. Therefore, it is necessary to extract the region of interest (ROI) of the ascending aorta to reduce computational load and accelerate segmentation. Furthermore, extracting the ROI of the ascending aorta avoids the influence of surrounding tissues and organs such as the descending aorta on subsequent segmentation. Step S100 aims to extract the ROI of the ascending aorta. Since the lung region has very obvious pixel features (i.e., CT values), the lung region can be extracted based on the CT values, and then the ROI of the ascending aorta can be defined using the lung region.

[0053] For details, please see the appendix. Figure 2 5:

[0054] S101. Extract the lung region from the coronary CTA image;

[0055] S102. Calculate the centroid of the lung region based on CT values;

[0056] S103. Define the region of interest of the ascending aorta within a preset range around the centroid.

[0057] Next, in the cross-sectional image of the coronary CTA, the ascending aorta is roughly circular. Utilizing this morphological feature, in step S200, within the region of interest of the ascending aorta, to avoid missing possible locations of the ascending aorta, Hough circle detection is used to calculate the center coordinates and radii of three circles that might contain the ascending aorta. These extracted circular regions are then filtered using the feature parameters of the ascending aorta. After selecting the circular region most likely to be the ascending aorta, leveraging the continuity of the ascending aorta, the center coordinates and radius of the largest circular region are selected as the starting position coordinates and radius of the ascending aorta using the largest connected component labeling.

[0058] For details, please see the appendix. Figure 3 6:

[0059] S201. Calculate n first-class detection circles that may contain the ascending aorta region using the Hough circle detection method, and determine the center coordinates and radius of the detection circles; n≥3;

[0060] S202. Select the second type of detection circle that is most likely to be the ascending aorta from the first type of detection circle based on the characteristic parameters of the ascending aorta;

[0061] S203. Using the maximum connected component label, select the center coordinates and radius of the detection circle with the largest area in the second type of detection circle as the starting position center coordinates and starting position radius of the ascending aorta.

[0062] Among them, step S202, which selects the second type of detection circle most likely to be the ascending aorta from the first type of detection circle based on the characteristic parameters of the ascending aorta, includes:

[0063] (1) Utilizing the morphological feature of the ascending aorta being roughly circular, select the detection circles in the first type of detection circles whose cumulative weight of Hough is greater than or equal to the preset weight value;

[0064] (2) Using the CT value characteristics of the ascending aorta in CTA images, select the detection circle in the first type of detection circle whose average pixel gray value is greater than or equal to the preset pixel gray value;

[0065] (3) Taking advantage of the uniform distribution of CT values ​​in the ascending aorta, select the detection circle in the first type of detection circle where the standard deviation of pixel gray level is greater than or equal to the preset standard deviation.

[0066] Next, after selecting the center coordinates and radius of the detection circle with the largest area in the second type of detection circle as the starting position center coordinates and starting position radius of the ascending aorta, it is also necessary to determine the position coordinates and radius of the ascending aorta in each cross section based on the starting position coordinates and radius of the ascending aorta.

[0067] For details, please see the appendix. Figure 3 :

[0068] S204. Starting from the center coordinates of the initial position of the ascending aorta, calculate the gradient energy of each cross section of the ascending aorta.

[0069] S205. Within the first preset neighborhood of the center coordinates of the starting position of the ascending aorta, and within the second preset neighborhood of the radius of the starting position, the coordinates corresponding to the maximum gradient energy are taken as the center coordinates of the cross section of the ascending aorta, and the radius corresponding to the maximum gradient energy is taken as the radius of each cross section of the ascending aorta.

[0070] Specifically, step S204 calculates the energy of each cross-section of the ascending aorta based on the gradient energy corresponding to the i-th pixel. GE xi Let be the gradient value of the i-th pixel in the x-direction; GE yi Let be the gradient value of the i-th pixel in the y-direction; GE zi Let be the gradient value of the i-th pixel in the z-direction.

[0071] Finally, the ascending aorta is segmented using the GraphCut method optimized in this invention. See Appendix. Figure 4 Specifically, it includes:

[0072] S301. Determine the initial ascending aorta mask based on the location coordinates and radius of the ascending aorta;

[0073] S302. Erode the initial ascending aorta mask to create a foremask; dilate the initial ascending aorta mask with varying degrees of intensity to create a backmask; segment the initial location region of the ascending aorta using a plotting method.

[0074] S303, Calculate the weight energy CapSourse[x] of the foreground mask. i y i , z i The weight energy of the background mask, CapSink[x] i y i , z i ];

[0075] S304. Calculate energy E using the energy function;

[0076] S305, Select the minimum energy value min(E[x]). i y i , z i The segmentation result corresponding to ]);

[0077] S306. Continuously optimize the foreground and background masks, and use the image cutting method to segment the initial location region of the ascending aorta until the rate of change of the minimum energy value between two adjacent segments is less than 0.1.

[0078] S307. Smooth the edges of the segmentation results to obtain the ascending aorta segmentation results.

[0079] Step S303 calculates the weight energy CapSourse[x] of the foreground mask. i y i , z i The weight energy of the background mask, CapSink[x] i y i, z i The steps include: (1) If the value of the foreground mask of the i-th pixel is 1, then the weight energy of the foreground mask of the i-th pixel is the maximum value; if the value of the background mask of the i-th pixel is 1, then the weight energy of the foreground mask of the i-th pixel is 0; otherwise, the weight energy of the foreground mask of the i-th pixel is p[ImageDate[x] i y i , z i (2) If the value of the background mask of the i-th pixel is 1, then the weight energy of the background mask of the i-th pixel is the maximum value; otherwise, the weight energy of the background mask of the i-th pixel is 0. The specific steps are as follows.

[0080] First, calculate the grayscale histogram of the image in the foreground mask, and then count the frequency p of each pixel grayscale value in the foreground mask.

[0081] Next, the weight energy CapSourse[x] of the foreground mask is calculated. i y i , z i ]:

[0082] If ForeMask[x] i y i , z i If ] = 1, then CapSourse[x] i y i , z i ] = MaxValue;

[0083] If BackMask[x i y i , z i If ] = 1, then CapSourse[x] i y i , z i ] = 0;

[0084] Otherwise, CapSourse[x i y i , z i ] = p[ImageDate[x i y i , z i ]]; where, [ImageDate[x i y i , z i ]] represents the grayscale value of the i-th pixel.

[0085] Finally, the weight energy of the background mask, CapSink[x], is calculated. i y i, z i ]:

[0086] If BackMask[x i y i , z i If ] = 1, then CapSink[x i y i , z i ] = MaxValue;

[0087] Otherwise, CapSink[x i y i , z i ] = 0.

[0088] Furthermore, the weight energy CapSourse[x] of the foreground mask is calculated. i y i , z i The weight energy of the background mask, CapSink[x] i y i , z i Next, the weighted energy of the 6-neighborhood of all pixels needs to be calculated. The weighted energy of the 6-neighborhood of all pixels is denoted as CapLee, CapGee, CapEle, CapEge, CapEel, and CapEeg, respectively. Taking CapLee as an example, the weighted energy is calculated as follows:

[0089] If ForeMask[x] i y i , z i ] = 1 & ForeMask[x i -1, y i , z i ] = 1, or BackMask[x i y i , z i ] = 1 & BackMask[x i -1, y i , z i If ] = 1, then CapLee[x] i y i , z i ] = MaxValue;

[0090] otherwise,

[0091] Where MaxValue is the maximum value, (x i y i , z i ) represents the position coordinates of the i-th pixel, [ImageDate[x iy i , z i ]] represents the grayscale value of the i-th pixel.

[0092] Similarly, the weight energy of other neighborhoods (CapGee, CapEle, CapEge, CapEel, CapEeg) of the 6-neighborhood can be calculated.

[0093] The energy function of step S304 is: Energy E[x] i y i , z i ]=λR[x i y i , z i ]+B[x i y i , z i ].

[0094] Wherein, R[x i y i , z i R[x] represents the sum of the weighted energies of the foreground and background masks. i y i , z i ] = CapSourse[x i y i , z i ]+CapSink[x i y i , z i ];B[x i y i , z i [This represents the sum of weighted energy among 6 neighboring pixels.]

[0095] B[x i y i , z i ] = CapLee[x i y i , z i ]+CapGee[x i y i , z i ]+...+CapEeg[x i y i , z i ]; λ is the region

[0096] The influence factors of the weight energy of the domain and the weight energy between adjacent pixels.

[0097] Referring to Figures 7-9, examples of ascending aorta segmentation based on coronary CTA images are shown. (a) represents one group, and (b) represents another. Figures 8(a) and 9(a) are both results of segmentation based on the coronary CTA image in Figure 7(a). The ascending aorta segmented using the method of this invention in Figure 9(a) is significantly clearer than the segmentation result in Figure 8(a) using the prior art. Figures 8(b) and 9(b) are both results of segmentation based on the coronary CTA image in Figure 7(b). The ascending aorta segmented using the method of this invention in Figure 9(b) is also significantly clearer than the segmentation result in Figure 8(b) using the prior art.

[0098] Referring to Figure 10, the optimized GraphCut method of this invention can effectively handle the blurred boundaries of the root of the ascending aorta in different sections of the ascending aorta, and the segmentation of the ascending aorta only takes about 2 seconds, with high segmentation efficiency.

[0099] The segmentation method of the present invention is not limited to the segmentation of the ascending aorta, but is also applicable to the segmentation of other tissues or organs in CT images.

[0100] It should be noted that the embodiments of the present invention have better implementability and are not intended to limit the present invention in any way. Any person skilled in the art may use the above-disclosed technical content to change or modify it into equivalent effective embodiments. However, any modifications or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for ascending aorta segmentation based on CTA images, characterized in that, include: The region of interest in the ascending aorta in the CTA image was determined based on the lung region; The initial location region of the ascending aorta is determined using the Hough circle detection method, and the initial location region includes the center position coordinates and radius; The initial location region of the ascending aorta is subjected to erosion and expansion operations of varying intensities, serving as the foreground and background masks for the image tiling method. The initial location region of the ascending aorta is segmented using the graph cutting method, and the segmentation result corresponding to the shortest path is retained as the segmentation result of the ascending aorta. The method of determining the region of interest (ROI) of the ascending aorta in the CTA image based on the lung region includes: Extract the lung region from the CTA image; The centroid of the lung region is calculated based on CT values; The region within a preset range surrounding the centroid is defined as the region of interest of the ascending aorta; The determination of the initial location region of the ascending aorta using the Hough circle detection method includes: Calculate using the Hough circle detection method A first type of detection circle that may contain the ascending aortic region is defined, and the center coordinates and radius of the detection circle are determined. ; The second type of detection circle, which is most likely to be the ascending aorta, is selected from the first type of detection circles based on the characteristic parameters of the ascending aorta. Using the maximum connected component label, the center coordinates and radius of the detection circle with the largest area in the second type of detection circle are selected as the starting position center coordinates and starting position radius of the ascending aorta; After selecting the center coordinates and radius of the detection circle with the largest area in the second type of detection circle as the starting position center coordinates and starting position radius of the ascending aorta, the method further includes: Starting from the center coordinates of the initial position of the ascending aorta, calculate the gradient energy of each cross section of the ascending aorta; Within a first preset neighborhood of the center coordinates of the starting position of the ascending aorta, and within a second preset neighborhood of the radius of the starting position, the coordinates corresponding to the maximum gradient energy are taken as the center coordinates of the cross section of the ascending aorta, and the radius corresponding to the maximum gradient energy is taken as the radius of each cross section of the ascending aorta. The calculation of the energy of each cross-section of the ascending aorta includes: No. Gradient energy corresponding to each pixel ; in, For the first Pixels in Gradient value in the direction; For the first Pixels in The gradient value in the direction; For the first Pixels in The gradient value in the direction.

2. The method for ascending aorta segmentation based on CTA images according to claim 1, characterized in that, The step of selecting the second type of detection circle, which is most likely to be the ascending aorta, from the first type of detection circle based on the characteristic parameters of the ascending aorta includes: The detection circles in the first type of detection circles whose cumulative Hough weight is greater than or equal to a preset weight value are selected; the detection circles in the first type of detection circles whose average pixel gray value is greater than or equal to a preset pixel gray value are selected; and the detection circles in the first type of detection circles whose standard deviation of pixel gray value is greater than or equal to a preset standard deviation are selected, thereby selecting the second type of detection circle that is most likely to be the ascending aorta among the first type of detection circles.

3. The method for ascending aorta segmentation based on CTA images according to claim 1, characterized in that, The initial location region of the ascending aorta is subjected to erosion and expansion operations of varying intensities, serving as the foreground and background masks for the image tiling method. The initial location region of the ascending aorta is segmented using a graph cutting method, and the segmentation result corresponding to the shortest path is retained as the segmentation result of the ascending aorta, including: Determine the initial ascending aortic mask based on the location coordinates and radius of the ascending aorta; The initial ascending aortic mask is etched and used as a foreground mask. The initial ascending aortic mask was expanded at different intensities to serve as a background mask. The initial location region of the ascending aorta is segmented using the graphical method; Calculate the weight energy of the foreground mask. and the weight energy of the background mask ; Energy through energy function Calculation; Select the minimum energy value The corresponding segmentation results; The foreground and background masks are continuously optimized, and the initial location region of the ascending aorta is segmented using the graph cutting method until the rate of change of the minimum energy value between two adjacent segments is less than 0.

1. Edge smoothing is performed on the segmentation results to obtain the ascending aorta segmentation results.

4. The method for ascending aorta segmentation based on CTA images according to claim 3, characterized in that, The calculation of the weight energy of the foreground mask and the weight energy of the background mask include: Calculate the gray-level histogram of the image in the foreground mask, and count the frequency of each pixel's gray-level value in the foreground mask. ; like ,but ,in It is the maximum value; like ,but ; otherwise, ; For the first The pixel grayscale value corresponding to each pixel; like ,but ; otherwise, .

5. The method for ascending aorta segmentation based on CTA images according to claim 3, characterized in that, The calculation of the weight energy of the foreground mask and the weight energy of the background mask It also includes calculating the weighted energy of the 6-neighborhood of all pixels: The weight energy of one neighborhood for: like , or , but ; otherwise, ; in, The maximum value, For the first The position coordinates of each pixel. For the first The pixel grayscale value corresponding to each pixel. denoted as the standard deviation of the Gaussian weighting function.

6. The method for ascending aorta segmentation based on CTA images according to claim 3, characterized in that, The energy is obtained through the energy function. The calculations include: energy ; in, The sum of the weighted energies of the foreground and background masks. It is the sum of the weighted energy among the 6 neighboring pixels. This is the influence factor of the weight energy of the region and the weight energy between adjacent pixels.

Images