A software-based method for IV-OCT image vessel wall segmentation
Through Gabor filtering and polar coordinate conversion methods, the problem of difficulty in segmenting blood vessel wall contour in IV-OCT technology is solved, and accurate tube wall contour segmentation of IV-OCT images is achieved, improving the resolution and accuracy of imaging.
Patent Information
- Application Number
- CN202210075802.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-23
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-01-23
AI Technical Summary
The existing IV-OCT technology is interfered by factors such as red blood cells, red thrombus and lipid core during observation of blood vessel wall structure and estimation of plaque load. The imaging resolution is limited, making it difficult to accurately segment the contour of blood vessel wall.
The software-level Gabor filtering processing method is used to detect the texture edge of the IV-OCT image. By generating Gabor filter cores at different angles, convolution operation is performed, combined with polar coordinate conversion and threshold segmentation, fully automatic segmentation of blood vessel wall profile is achieved.
Accurate tube wall contour segmentation of IV-OCT images is achieved, improving imaging resolution and accuracy, and is suitable for imaging healthy and lesional blood vessels.
Smart Images

Figure CN114494178B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of OCT image processing, and specifically relates to a method for processing IV-OCT images using Gabor filtering at a software level, and performing tube wall contour segmentation on the processed images. Background Art
[0002] Optical Coherence Tomography (OCT) is a non-invasive detection technology. OCT technology has been widely used in in vivo cross-sectional structural imaging of biological tissues. By measuring the scattered light related to depth, OCT can provide high-resolution and high-sensitivity tissue structures.
[0003] When OCT is applied to vascular imaging (IV-OCT), a rotatable optical lens and optical fiber are used to emit near-infrared light to the inner surface of the blood vessel, and an optical interferometer is used to receive the reflected light waves and form an image. Because it uses light waves for imaging, OCT imaging has high resolution, and the axial resolution can reach 10-20um, which can image the composition and microstructure of the plaque surface. However, near-infrared light waves do not have strong penetration (about 1.0-2.5mm), and blood cells, red thrombi, and the lipid core of the plaque or necrosis within the plaque will affect OCT's observation of the vascular wall structure and the estimation of plaque load. Because red blood cells scatter red light, OCT in the past needed to continuously inject contrast agents to flush blood away during imaging. Modern OCT systems have partially reduced the interference of red blood cells and other factors on imaging through technologies such as rapid rotation and retraction, and can complete the imaging of a length of blood vessels in a few seconds.
[0004] The Gabor filter is a linear filter used for edge detection. The frequency and direction representation of the Gabor filter is close to the representation of frequency and direction by the human visual system, and they are often used for texture representation and description. In the spatial domain, a 2D Gabor filter is a product of a sinusoidal plane wave and a Gaussian kernel function. It has the characteristics of achieving optimal localization in both the spatial domain and the frequency domain, which is very similar to the biological vision characteristics of humans. Therefore, it can well describe the local structural information corresponding to spatial frequency (scale), spatial position and direction selectivity.
[0005] For healthy IV-OCT images, blood vessels show a three-layer membrane structure, including the inner membrane, the middle membrane, and the outer membrane. The inner membrane is composed of a single layer of endothelial cells, extracellular matrix, and fibers, and appears as a bright and uniform thin layer structure in the IV-OCT image. The middle membrane is sandwiched between the inner and outer membranes, composed of smooth muscle cells and extracellular matrix, and appears as a uniform dark band in the IV-OCT image. The outer membrane also appears as a bright and uniform thin layer structure due to its rich fibrous tissue. The area outside the outer membrane is darker and has a distinct texture, with a clear boundary with the outer membrane. Summary of the invention
[0006] The present invention discloses a software-level IV-OCT image tube wall segmentation method, which aims to find and automatically segment the IV-OCT tube wall contour.
[0007] The technical solution of the present invention is as follows:
[0008] A software-level IV-OCT image vessel wall segmentation method, characterized by comprising the following steps:
[0009] S101 reads the IV-OCT image, obtains the image resolution f, and uses a certain blood vessel segmentation method to obtain the vascular lumen contour information based on the image
[0010] S102: If the input OCT image is an RGB image, convert it into a grayscale image (the range of the pixel value of each point is 0-255).
[0011] S103 performs Gabor filtering on the grayscale image of IV-OCT, generates a corresponding filter kernel using the Gabor function, and performs a convolution operation on the grayscale IV-OCT image using the filter kernel. The mathematical expression of the two-dimensional Gabor function is as follows:
[0012] Plural expression:
[0013] Real part:
[0014] Imaginary part: Where: x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ. Wavelength (λ): Its value is specified in pixels, usually greater than or equal to 2. But it cannot be greater than one-fifth of the input image size. In this method, PI=3.14159; Direction (θ): This parameter specifies the direction of the parallel stripes of the Gabor function, and its value ranges from 0 to 360 degrees; Phase offset Its value range is -180 degrees to 180 degrees. Among them, 0 and 180 degrees correspond to the center-on function and center-off function of the center symmetry, respectively, while -90 degrees and 90 degrees correspond to the antisymmetric function. In this method, 0° is taken; aspect ratio (γ): spatial aspect ratio, which determines the ellipticity of the Gabor function shape (support, I translate it as shape). When γ = 1, the shape is circular. When γ < 1, the shape is elongated along the direction of parallel stripes. In this method, this value is taken as 0.5; bandwidth (b): the half-response spatial frequency bandwidth b of the Gabor filter is related to the ratio of σ / λ, where σ represents the standard deviation of the Gabor function's Gaussian factor: The value of σ cannot be set directly, it only changes with the bandwidth b. The bandwidth value must be a positive real number, usually 1, in which case the relationship between the standard deviation and the wavelength is: σ = 0.56λ. The smaller the bandwidth, the larger the standard deviation, the larger the Gabor shape, and the more parallel excitation and inhibition zone stripes are visible.
[0015] Furthermore, the size of the filter kernel is selected to be 7*7 pixels, and the filter angles are selected to be 15°~150° with an interval of 15°. One frame of IV-OCT image obtains 12 Gabor filter processing result images at different angles. By selecting different filter angles, textures at different angles can be screened out, and finally they can be spliced together to obtain textures covering most angles.
[0016] S104 further combines the 12 images into a filtered result image, and the combining method selects the maximum value of the same position in the 12 images for each pixel: Among them I ij is the pixel value of the i-th row and j-th column, is the pixel value of the i-th row and j-th column of the filtered result image at the n-th angle.
[0017] S105 performs polar coordinate conversion, and obtains the distance Rθ from the lumen contour to the center point at each angle θ within the range of 0-360° according to the lumen contour information.
[0018] S106 further finds a ray extending from the center point to the outside of the blood vessel at each angle. For each angle θ, the ray is started from a point Rθ away from the center point (i.e., the lumen contour point) and the step is 1 pixel. The following operations are performed in a loop:
[0019] S107 In the i-th cycle, the distance R from the point to the center point is Rθ+i, and the sum S of the pixel values of the merged filtering result I at the corresponding position in the circular area A with a radius of r (here r=10) pixels with the point as the center is calculated: Where (x0, y0) is the center point of the image. The coordinates of the contour point in the rectangular coordinate system are obtained according to the distance R to the center point and the angle θ, and then rounded to (x, y). The point set of the circular area with (x, y) as the center is A = {(x n ,y n )|(x n -x) 2 +(y n -y) 2 ≤r;x n ,y n ∈Z}, the sum of the pixel values of image I in the region is
[0020] In the above loop, when the pixel value and S are greater than a certain threshold S>ST (here ST=3500) for the first time, the loop ends, and the distance R obtained in the current loop is selected to calculate the distance from the boundary point of the tube wall to the center point: R'=R+r, and then the rectangular coordinates (x', y') are calculated: This point is the corresponding lumen contour point at the angle θ.
[0021] S108, finding the coordinates of the contour points at each angle θ, and connecting the points adjacent to each other in the angle order to obtain the tube wall contour segmentation result of the IV-OCT image.
[0022] Compared with the prior art, the advantages of the present invention are as follows: based on the image characteristics of healthy IV-OCT, Gabor filtering is used for texture edge detection to achieve accurate IV-OCT management contour segmentation; different filtering angles are spliced together to screen out the final filtering result containing most angle textures; based on the imaging characteristics of IV-OCT, threshold segmentation is performed in the form of polar coordinates based on the filtering results, and this method has good versatility. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 It is the original IV-OCT image;
[0024] Figure 2 is a flow chart of the present invention;
[0025] Figure 3 This is the Gabor filtering result at an angle of 0°;
[0026] Figure 4 The Gabor filter results at 12 angles are merged;
[0027] Figure 5 This is the final tube wall contour segmentation result. DETAILED DESCRIPTION
[0028] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0029] refer to Figure 2 , Figure 2 The flowchart shown in the present invention comprises the following steps:
[0030] The technical solution of the present invention is as follows:
[0031] The present invention discloses a software-level IV-OCT image tube wall segmentation method, which aims to find and automatically segment the IV-OCT tube wall contour.
[0032] The technical solution of the present invention is as follows:
[0033] Embodiment 1: A software-level IV-OCT image vessel wall segmentation method comprises the following steps:
[0034] S101 reads the IV-OCT image, obtains the image resolution f, and uses a certain blood vessel segmentation method to obtain the vascular lumen contour information based on the image
[0035] S102: If the input OCT image is an RGB image, convert it into a grayscale image (the range of the pixel value of each point is 0-255).
[0036] S103 performs Gabor filtering on the grayscale image of IV-OCT, generates a corresponding filter kernel using the Gabor function, and performs a convolution operation on the grayscale IV-OCT image using the filter kernel. The mathematical expression of the two-dimensional Gabor function is as follows:
[0037] Plural expression:
[0038] Real part:
[0039] Imaginary part: Where: x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ. Wavelength (λ): Its value is specified in pixels, usually greater than or equal to 2. But it cannot be greater than one-fifth of the input image size. In this method, PI=3.14159; Direction (θ): This parameter specifies the direction of the parallel stripes of the Gabor function, and its value ranges from 0 to 360 degrees; Phase offset Its value range is -180 degrees to 180 degrees. Among them, 0 and 180 degrees correspond to the center-on function and center-off function of the center symmetry, respectively, while -90 degrees and 90 degrees correspond to the antisymmetric function. In this method, 0° is taken; aspect ratio (γ): spatial aspect ratio, which determines the ellipticity of the Gabor function shape (support, I translate it as shape). When γ = 1, the shape is circular. When γ < 1, the shape is elongated along the direction of parallel stripes. In this method, this value is taken as 0.5; bandwidth (b): the half-response spatial frequency bandwidth b of the Gabor filter is related to the ratio of σ / λ, where σ represents the standard deviation of the Gabor function's Gaussian factor: The value of σ cannot be set directly, it only changes with the bandwidth b. The bandwidth value must be a positive real number, usually 1, in which case the relationship between the standard deviation and the wavelength is: σ = 0.56λ. The smaller the bandwidth, the larger the standard deviation, the larger the Gabor shape, and the more parallel excitation and inhibition zone stripes are visible.
[0040] For step S103, the filter kernel size is selected as 7*7 pixels, and the filter angles are selected as 15° to 150° with an interval of 15°. One frame of IV-OCT image obtains 12 Gabor filter processing result images at different angles. The merged image results are referenced Figure 4 .
[0041] S104 combines the 12 images into a filtering result image, and the combining method selects the maximum value of the same position in the 12 images for each pixel:
[0042] Among them I ij is the pixel value of the i-th row and j-th column, is the pixel value of the i-th row and j-th column of the filtered result image at the n-th angle.
[0043] S105 performs polar coordinate conversion, and obtains the distance Rθ from the lumen contour to the center point at each angle θ within the range of 0-360° according to the lumen contour information.
[0044] S106 finds a ray extending from the center point to the outside of the blood vessel at each angle. For each angle θ, the ray is started from a point Rθ away from the center point (i.e., the lumen contour point) and the step is 1 pixel. The following operations are performed in a loop:
[0045] S107 In the i-th cycle, the distance R from the point to the center point is Rθ+i, and the sum S of the pixel values of the merged filtering result I at the corresponding position in the circular area A with a radius of r (here r=10) pixels with the point as the center is calculated: Where (x0, y0) is the center point of the image. The coordinates of the contour point in the rectangular coordinate system are obtained according to the distance R to the center point and the angle θ, and then rounded to (x, y). The point set of the circular area with (x, y) as the center is A = {(x n ,y n )|(x n -x) 2 +(y n -y) 2 ≤r;x n ,y n ∈Z}, the sum of the pixel values of image I in the region is
[0046] In the loop of step S107, when the pixel value and S are greater than a certain threshold value S>ST (here ST=3500) for the first time, the loop ends, and the distance R obtained in the current loop is selected to calculate the distance from the boundary point of the pipe wall to the center point: R'=R+r, and then the rectangular coordinates (x', y') are calculated: This point is the corresponding lumen contour point at the angle θ.
[0047] S108 finds the coordinates of the contour points at each angle θ, and connects the adjacent points in the angle order to obtain the tube wall contour segmentation result of the IV-OCT image. Figure 5 .
[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention. The present invention is further explained below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.
Claims
1. A software-based IV-OCT image vessel wall segmentation method, characterized in that: It includes the following 8 steps: S101 reads the IV-OCT image, obtains the image resolution f, and obtains the vascular lumen contour information according to the image using a certain vascular segmentation method; S102: if the input OCT image is an RGB image, convert it into a grayscale image; S103 performs Gabor filtering on the grayscale image of IV-OCT, generates a corresponding filter kernel using a Gabor function, and performs a convolution operation on the grayscale IV-OCT image using the filter kernel; S104 combines the 12 images into a filtering result image, and the combining method selects the maximum value of the same position in the 12 images for each pixel: Among them I ij is the pixel value of the i-th row and j-th column, is the pixel value of the i-th row and j-th column of the filtered result image at the n-th angle; S105 performs polar coordinate conversion to obtain the distance Rθ from the lumen contour to the center point at each angle θ within a range of 0-360° according to the lumen contour information; S106 finds a ray extending from the center point to the outside of the blood vessel at each angle. For each angle θ, the ray is started from a point Rθ away from the center point, i.e., the lumen contour point, with a step of 1 pixel, and the following operations are performed in a loop: S107 In the i-th cycle, the distance R from the point to the center point is Rθ+i, and the sum S of the pixel values of the merged filtering result I at the corresponding positions in the circular area A of the pixel with the radius r and r=10 is calculated: Where (x0, y0) is the center point of the image. The coordinates of the contour point in the rectangular coordinate system are obtained according to the distance R to the center point and the angle θ, and then rounded to (x, y). The point set of the circular area with (x, y) as the center is A = {(x n ,y n )|(x n -x) 2 +(y n -y) 2 ≤r;x n ,y n ∈Z}, the sum of the pixel values of image I in the region is S108 finds the coordinates of the contour points at each angle θ, and connects the points adjacent to each other in the angle order to obtain the tube wall contour segmentation result of the IV-OCT image.
2. A software-level IV-OCT image vessel wall segmentation method according to claim 1, characterized in that: In step S103, the mathematical expression of the two-dimensional Gabor function is as follows: Plural expression: Real part: Imaginary part: Where: x′=x cosθ+ysinθ, y′=-x sinθ+y cosθ, wavelength λ: its value is specified in pixels, greater than or equal to 2, and cannot be greater than one-fifth of the input image size, take PI=3.14159; direction θ: this parameter specifies the direction of the parallel stripes of the Gabor function, and its value ranges from 0 to 360 degrees; phase offset Its value range is -180 degrees to 180 degrees, where 0 and 180 degrees correspond to the center-on function and center-off function of the center symmetry, respectively, while -90 degrees and 90 degrees correspond to the antisymmetric function, and 0° is taken; Aspect ratio γ: spatial aspect ratio, which determines the ellipticity of the Gabor function shape. When γ = 1, the shape is circular, and when γ < 1, the shape is elongated along the direction of parallel stripes. The value is 0.5; Bandwidth b: The half-response spatial frequency bandwidth b of the Gabor filter is related to the ratio of σ / λ, where σ represents the standard deviation of the Gabor function's Gaussian factor.
3. The software-level IV-OCT image vessel wall segmentation method according to claim 1, characterized in that: In step S103, the size of the filter kernel is selected to be 7*7 pixels, and the filter angles are selected to be 15° to 150° with an interval of 15°. One frame of IV-OCT image obtains 12 Gabor filter processing result images at different angles.
4. The software-level IV-OCT image vessel wall segmentation method according to claim 1, characterized in that: In the loop of step S107, when the pixel value and S are greater than a certain threshold value S>ST for the first time, ST=3500, the loop ends, and the distance R obtained in the current loop is selected to calculate the distance from the boundary point of the pipe wall to the center point: R'=R+r, and then the rectangular coordinates (x', y') are calculated: This point is the corresponding lumen contour point at the angle θ.
Citation Information
Patent Citations
IVOCT image vessel wall lumen contour extraction method
CN109118508A
IV-OCT blood vessel contour recognition method based on uniqueness of tube wall connected domain
CN112308874A
System and Method For Statistical Shape Model Based Segmentation of Intravascular Ultrasound and Optical Coherence Tomography Images
US20080075375A1
