A method and device for extracting blood vessels in an OCT fundus image

Abstract

The application provides a blood vessel extraction method and device in an OCT fundus image, and the method comprises the following steps: pre-processing the OCT fundus image to obtain a first process image, wherein the pre-processing comprises automatic positioning and removal of a lesion area; extracting global features of the first process image through matched filtering to separate out a suspected blood vessel area and obtain a second process image; removing interference features in the second process image through a pore filling method based on a capillary bundle model to obtain a third process image; and performing threshold cutting processing and morphological processing on the third process image to obtain a final blood vessel image. The application can effectively remove interference such as blood vessel lesions and accurately extract clear blood vessel features, thereby greatly improving the extraction effect of blood vessels in the OCT fundus image.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and specifically to a method and apparatus for extracting blood vessels from OCT fundus images. Background Technology

[0002] In ophthalmology, because the choroid lacks sensory nerves, early-stage subretinal neovascularization is asymptomatic and painless. As it gradually enlarges, it can cause hemorrhages and leaks in the fovea, significantly obstructing central vision and impairing central visual field. It is common in adult eyes, especially those over 60 years of age, and early detection and timely treatment are crucial to controlling its development. Using OCT (Optical Coherence Tomography) imaging at multiple time points, the condition can be predicted, guiding treatment planning. The first step in predicting choroidal neovascularization (CNV) growth is registering OCT data from different time points. Retinal vascular information in OCT images is a key feature for registration; therefore, extracting vascular features from OCT fundus images has significant clinical implications and is a hot research topic in the field of medical engineering.

[0003] However, unlike color fundus photography, OCT en-face fundus images present the following challenges: 1) The grayscale values ​​of vascular features are close to the background and are accompanied by noise; 2) During image acquisition, discontinuous vascular shadows or double images may occur due to patient eye movements; 3) Especially for fundus vascular lesions, symptoms such as fluid accumulation and hemorrhage caused by CNVs can interfere with normal effective vascular features, and the lesion vessels have similar features to normal vessels. During extraction, the lesion vessels will be extracted along with the normal vessels, which will interfere with subsequent steps, making the segmentation problem more complex.

[0004] Current solutions are mostly based on color fundus images, and most of them are based on normal retinal images. Although there are solutions for lesions, there are still few solutions for CNV lesions. In addition, due to the characteristics of en-face images themselves, existing solutions are difficult to accurately extract blood vessels. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and apparatus for extracting blood vessels from OCT fundus images. This method effectively removes interference such as vascular lesions and accurately extracts clear vascular features, significantly improving the extraction effect of blood vessels from OCT fundus images.

[0006] The technical solution adopted in this invention is as follows:

[0007] A method for extracting blood vessels from OCT fundus images includes the following steps: preprocessing the OCT fundus image to obtain a first process image, wherein the preprocessing includes automatic localization and removal of lesion areas; extracting global features from the first process image using matched filtering to separate suspected blood vessel areas, thereby obtaining a second process image; removing interfering features from the second process image using a pore-filling method based on a capillary bundle model, thereby obtaining a third process image; and performing threshold truncation and morphological processing on the third process image to obtain a final blood vessel image.

[0008] The preprocessing of the OCT fundus image specifically includes: performing median filtering on the OCT fundus image; sharpening the median-filtered image using the Laplacian operator; automatically locating the lesion region in the sharpened image through grayscale calculation; and performing a coverage process on the lesion region to remove the lesion region.

[0009] The first process image is extracted using matched filtering, specifically by constructing a blood vessel extraction model based on a two-dimensional Gaussian kernel function, extracting suspected blood vessel features from the first process image using the blood vessel extraction model, separating the suspected blood vessel region from the background region, and obtaining the second process image.

[0010] The method of removing interference features in the second process image by pore filling based on capillary bundle model specifically includes: determining a point with a gray value of 0 in the pore space, and drawing an expanding circle outward from the point until the circle touches a point in the pore space with a gray value less than a preset threshold, at which point the expansion stops; storing the obtained circles in an array and arranging them in ascending order of radius; sequentially taking out one circle from the array and deleting the circles that intersect with or are contained within the circle until the last element; and setting the gray value of the region containing each remaining circle in the array to 0.

[0011] Morphological processing of the third process image specifically includes performing dilation and erosion operations on the third process image.

[0012] A device for extracting blood vessels from OCT fundus images includes: a preprocessing module for preprocessing the OCT fundus images to obtain a first process image, wherein the preprocessing includes automatic localization and removal of lesion areas; an extraction module for extracting global features from the first process image using matched filtering to separate suspected blood vessel areas to obtain a second process image; a removal module for removing interfering features from the second process image using a pore-filling method based on a capillary bundle model to obtain a third process image; and a post-processing module for performing threshold truncation and morphological processing on the third process image to obtain a final blood vessel image.

[0013] The preprocessing module is specifically used for: performing median filtering on the OCT fundus image; sharpening the median-filtered image using the Laplacian operator; automatically locating the lesion region in the sharpened image through grayscale calculation; and performing coverage processing on the lesion region to remove the lesion region.

[0014] The extraction module is specifically used to: construct a blood vessel extraction model based on a two-dimensional Gaussian kernel function, extract suspected blood vessel features in the first process image through the blood vessel extraction model, separate the suspected blood vessel region from the background region, and obtain the second process image.

[0015] The removal module is specifically used for: determining a point in the pore space with a gray value of 0, and drawing a continuously expanding circle outward from that point until the circle encounters a point in the pore space with a gray value less than a preset threshold, at which point the expansion stops; storing the obtained circles in an array and arranging them in ascending order of radius; sequentially taking out one circle from the array and deleting circles in the array that intersect with or are contained by that circle, until the last element; and setting the gray value of the area containing each remaining circle in the array to 0.

[0016] The post-processing module is specifically used to perform dilatation and erosion operations on the third process image.

[0017] The beneficial effects of this invention are:

[0018] By sequentially preprocessing OCT fundus images, extracting global features, performing pore filling based on a capillary bundle model, thresholding, and morphological processing, interference such as vascular lesions can be effectively removed, and clear vascular features can be accurately extracted, greatly improving the extraction effect of blood vessels in OCT fundus images. Attached Figure Description

[0019] Figure 1 This is a flowchart of a method for extracting blood vessels from OCT fundus images according to an embodiment of the present invention;

[0020] Figure 2 This is an image after median filtering and sharpening according to an embodiment of the present invention;

[0021] Figure 3 This is a first process image according to an embodiment of the present invention;

[0022] Figure 4 This is a second process image according to an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the pore space according to an embodiment of the present invention;

[0024] Figure 6This is a schematic diagram of a capillary bundle model according to an embodiment of the present invention;

[0025] Figure 7 This is a schematic diagram of the filling process of a pore filling method based on a capillary bundle model according to an embodiment of the present invention;

[0026] Figure 8 This is a third process image according to an embodiment of the present invention;

[0027] Figure 9 This is a final blood vessel image according to one embodiment of the present invention;

[0028] Figure 10 This is an OCT fundus image of a patient according to an embodiment of the present invention;

[0029] Figure 11(a) shows the effect of using a hybrid method based on the frangi method on... Figure 10 The final vascular image after vascular extraction;

[0030] Figure 11(b) shows the effect of multi-scale filtering on the filter. Figure 10 The final vascular image after vascular extraction;

[0031] Figure 11(c) shows the method of the present invention applied to... Figure 10 The final vascular image after vascular extraction;

[0032] Figure 12 This is a block diagram of a blood vessel extraction device in an OCT fundus image according to an embodiment of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] like Figure 1 As shown, the method for extracting blood vessels from OCT fundus images according to an embodiment of the present invention includes the following steps:

[0035] S1, preprocessing the OCT fundus image to obtain the first process image, wherein the preprocessing includes automatic localization and removal of lesion areas.

[0036] Specifically, the enface OCT fundus images can first be processed by median filtering to remove inherent noise from the images.

[0037] Then, the Laplacian operator is used to sharpen the median-filtered image to enhance the clarity and contrast of the blood vessel edges in the image.

[0038] In one embodiment of the present invention, the sharpening process can be accomplished through spatial differentiation, whereby the difference of the second derivative of a univariate function is defined as:

[0039]

[0040] The Laplace operator is defined as:

[0041]

[0042] The Laplacian operator is used to sharpen an image as follows:

[0043]

[0044] Where (x,y) represents pixel coordinates, f represents the input image, g represents the sharpened image, and δ is the template center coefficient. The sharpened image is obtained by adding the image calculated by the Laplacian operator to the original image, while preserving background information.

[0045] The image obtained after median filtering and sharpening of the enface OCT fundus image is shown below. Figure 2 As shown.

[0046] Furthermore, lesion areas in the sharpened image are automatically located through grayscale calculation. Specifically, a 3×3 rectangle can be used to traverse the image, and the threshold for region selection is determined by calculating the mean grayscale value of the rectangular region and the mean grayscale value of the image. The formula for calculating the mean grayscale value of the rectangular region and the mean grayscale value of the image is as follows:

[0047]

[0048]

[0049] Where, m r Let A(i,j) represent the average grayscale value of the rectangular region, (i,j) represent the coordinates of the pixel within the rectangle, and A(i,j) represent the grayscale value of the pixel within the rectangle. I Let m be the mean gray level of the image, M and N be the width and length of the image, respectively, and B(i,j) be the gray level value of a pixel in the image. The gray level interval X of the lesion is determined by the maximum inter-class variance method. If m r In the interval X, the x-coordinate of vertex 1 of the nth satisfying rectangle block is denoted as a, the x-coordinate of vertex 2 as b, the y-coordinate of vertex 4 as c, and the y-coordinate of vertex 1 as d. These are then placed into four arrays, denoted as P = [a1, a2, ..., a...]. n Q = [b1, b2, ..., b]n J = [c1, c2, ..., c n ], K = [d1, d2, ..., d n After the traversal is completed, the maximum value of arrays P and J and the minimum value of arrays Q and K are calculated respectively to determine the lesion area.

[0050] Finally, the lesion area is covered. For example, the gray average of the image can be used to cover the lesion area, that is, the gray values ​​of the lesion area are all replaced with the gray average of the image to remove the lesion area.

[0051] The first process image is as follows: Figure 3 As shown.

[0052] S2, global features are extracted from the first process image through matched filtering to separate the suspected blood vessel region and obtain the second process image.

[0053] Since the purpose of this invention is to extract blood vessels, that is, to explore vascular information, three representative features of blood vessels are selected to enhance the vascular feature information: 1) The curvature of blood vessels is small, and antiparallel pairs can be approximated by piecewise linear segments; 2) The reflectivity of blood vessels is lower than that of other surfaces, so they appear darker relative to the background; 3) Although the width of blood vessels decreases as they extend radially outward from the optic disc, this change in blood vessel diameter is gradual rather than abrupt.

[0054] Therefore, in one embodiment of the present invention, a blood vessel extraction model can be constructed based on a two-dimensional Gaussian kernel function to extract suspected blood vessel features in the first process image, separate the suspected blood vessel region from the background region, and obtain the second process image.

[0055] Specifically, the blood vessel extraction model is as follows:

[0056]

[0057] Among them, F θ σ is a two-dimensional Gaussian kernel function, where L represents the length of the segmented blood vessel and σ represents the blood vessel scale.

[0058] Since the direction of the selected blood vessel segments varies depending on the vessel's orientation, the Gaussian kernel function needs to be adjusted. This is achieved by rotating the kernel function to obtain templates with different orientations. Let A... i Let be the i-th rotation matrix that follows the deflection of the blood vessel, used to adjust the position of the pixel, and we have:

[0059]

[0060] Let f(x,y) be a point on the kernel function. After matrix operations, we obtain... but Given the rotated coordinates, the i-th matching function is:

[0061]

[0062] Where Z is the neighborhood of the template, and Z = {(u,v)||u|≤3δ,|v|≤L / 2}.

[0063] The second process image is as follows: Figure 4 As shown.

[0064] S3, interference features in the second process image are removed by a pore-filling method based on a capillary bundle model to obtain the third process image.

[0065] Due to the pathological characteristics of choroidal neovascularization (CNV), numerous proliferating vessels form between the choroid and the retinal pigment epithelium. These vessels are mostly short, densely packed, and granular, with a disordered and irregular arrangement between the various vessels originating from the lesion. "Follicular" structures form between these proliferating vessels. A capillary bundle model is introduced here to address these proliferating vessels. Figure 5 The diagram shows a pore space. In reality, pore spaces are irregular and need to be transformed into a series of regular capillaries. Assume the cross-section of each capillary is a regular circle. For example... Figure 6 Images show the process of extracting a capillary bundle model from an irregularly shaped porous region.

[0066] Specifically, a point with a gray value of 0 in the pore space can be identified, and a continuously expanding circle can be drawn outward from this point until the circle encounters a point in the pore space with a gray value less than a preset threshold, at which point the expansion stops. The resulting circles are stored in an array and arranged in ascending order of radius. One circle is taken out from the array in turn, and circles that intersect with or are contained by the circle are deleted until the last element. The gray value of the area containing each remaining circle in the array is set to 0.

[0067] In the pore-filling method of the capillary bundle model, it is assumed that the cross-section of the capillary bundle is circular. During the continuous outward expansion from non-particle pixels as the center, its radius is indeterminate until the circumference encounters a solid particle (in this invention, a proliferating blood vessel). This invention utilizes this characteristic, determining the condition for stopping expansion based on the pixel's grayscale value rather than "touching" it to generate all possible circles inside the pores, and assigning the background grayscale value of the circle to 0. The process is illustrated in the image below. Figure 7 As shown, circles that intersect or enclose other circles are deleted.

[0068] In fact, the initial purpose of the capillary bundle model is to transform irregular spatial shapes into a series of regular capillaries without affecting the particles (proliferating blood vessels) in the image. In this invention, the image obtained by matched filtering is first normalized, and the cutoff condition for circumferential expansion is reset according to the gray values. In order to retain more vascular information, 10% of the non-zero gray values ​​is used as the termination condition for circumferential expansion, that is, the preset threshold is the mean of 10% of non-zero gray values. According to the capillary bundle model, a portion of proliferating blood vessels (especially dense areas) can be largely ablated without interfering with the characteristics of the main blood vessels.

[0069] The specific calculation process is as follows:

[0070] Step 1: Determine the coordinates of the pixel with a grayscale value of 0. Using the point with a grayscale value of 0 as the center, record it as C. i Its corresponding coordinates are (x i ,y i (i = 1, 2, 3, ...).

[0071] Step 2: Calculate the average value v of all non-zero grayscale values. termination Gray values ​​less than 0.1×v termination The point record is D j Its corresponding coordinates are (x j ,y j (j = 1, 2, 3, ...). Calculate C. i To D j Distance:

[0072]

[0073] Then we have:

[0074]

[0075] Among them, R i Therefore, C i The radius of the circle with center is .

[0076] Step 3: Delete duplicate or unnecessary circles. For some intersecting circles or circles surrounded by other circles, they must be deleted. Store the circles in an array of centers and their corresponding radii obtained in Step 2, sorted by radius in ascending order. Then, calculate the distance from the center of the i-th circle to the center of the circle following it, in sorted order (i-1, i-2, ...).

[0077]

[0078] If there is dis <R i +R i+1 If so, then delete the current circle.

[0079] The non-intersecting circles from Step 3 are retained in the array and filled with the background color. These circles almost fill the entire pore space and eliminate interfering features that constitute the shape of the pores.

[0080] The third process image is as follows: Figure 8 As shown.

[0081] S4. Threshold truncation and morphological processing are performed on the third process image to obtain the final blood vessel image.

[0082] By selecting an appropriate threshold based on the mean, threshold truncation of the third-process image can further reduce interference from irrelevant features. Using a square structuring element with a parameter set to 2, dilation and erosion operations are performed on the threshold-truncated image to remove noise and pseudo-branches. The resulting final blood vessel image is shown below. Figure 9 As shown.

[0083] In one specific embodiment of the present invention, to verify the effectiveness of the method of the present invention, [the following is performed / conducted / etc.]. Figure 10 The enface OCT fundus images of patients with CNV disease shown were used for vessel extraction using a hybrid method based on the frangi method, a multi-scale filtering method, and the method of this embodiment, respectively, resulting in the corresponding... Figure 11(a) , 11(b) Final vascular images of 11(c). (From...) Figure 11(a) , 11(b) As can be seen from 11(c), the method of the present invention can effectively remove interference such as vascular lesions, accurately extract clear vascular features, and greatly improve the extraction effect of blood vessels in OCT fundus images.

[0084] Corresponding to the above-described method for extracting blood vessels from OCT fundus images, this invention also proposes a device for extracting blood vessels from OCT fundus images.

[0085] like Figure 12 As shown, the vessel extraction device in OCT fundus images according to an embodiment of the present invention includes a preprocessing module 10, an extraction module 20, a removal module 30, and a post-processing module 40. The preprocessing module 10 preprocesses the OCT fundus image to obtain a first process image, wherein the preprocessing includes automatic localization and removal of lesion areas. The extraction module 20 extracts global features from the first process image using matched filtering to separate suspected vascular regions, obtaining a second process image. The removal module 30 removes interfering features from the second process image using a pore-filling method based on a capillary bundle model, obtaining a third process image. The post-processing module 40 performs threshold truncation and morphological processing on the third process image to obtain the final vessel image.

[0086] In one embodiment of the present invention, the preprocessing module 10 is specifically used for: performing median filtering on the OCT fundus image; sharpening the median-filtered image using the Laplacian operator; automatically locating the lesion area in the sharpened image through grayscale calculation; and performing coverage processing on the lesion area to remove the lesion area.

[0087] In one embodiment of the present invention, the sharpening process can be accomplished through spatial differentiation, whereby the difference of the second derivative of a univariate function is defined as:

[0088]

[0089] The Laplace operator is defined as:

[0090]

[0091] The Laplacian operator is used to sharpen an image as follows:

[0092]

[0093] Where (x,y) represents pixel coordinates, f represents the input image, g represents the sharpened image, and δ is the template center coefficient. The sharpened image is obtained by adding the image calculated by the Laplacian operator to the original image, while preserving background information.

[0094] The image obtained after median filtering and sharpening of the enface OCT fundus image is shown below. Figure 2 As shown.

[0095] The preprocessing module 10 can use a 3×3 rectangle to traverse the image. It determines the threshold for region filtering by calculating the mean grayscale value of the rectangular region and the mean grayscale value of the image. The formula for calculating the mean grayscale value of the rectangular region and the image is as follows:

[0096]

[0097]

[0098] Where, m r Let A(i,j) represent the average grayscale value of the rectangular region, (i,j) represent the coordinates of the pixel within the rectangle, and A(i,j) represent the grayscale value of the pixel within the rectangle. I Let m be the mean gray level of the image, M and N be the width and length of the image, respectively, and B(i,j) be the gray level value of a pixel in the image. The gray level interval X of the lesion is determined by the maximum inter-class variance method. If m rIn the interval X, the x-coordinate of vertex 1 of the nth satisfying rectangle block is denoted as a, the x-coordinate of vertex 2 as b, the y-coordinate of vertex 4 as c, and the y-coordinate of vertex 1 as d. These are then placed into four arrays, denoted as P = [a1, a2, ..., a...]. n Q = [b1, b2, ..., b] n J = [c1, c2, ..., c n ], K = [d1, d2, ..., d n After the traversal is completed, the maximum value of arrays P and J and the minimum value of arrays Q and K are calculated respectively to determine the lesion area.

[0099] The preprocessing module 10 can use the grayscale mean of the image to cover the lesion area, that is, replace the grayscale value of the lesion area with the grayscale mean of the image to remove the lesion area.

[0100] The first process image is as follows: Figure 3 As shown.

[0101] In one embodiment of the present invention, the extraction module 20 can construct a blood vessel extraction model based on a two-dimensional Gaussian kernel function to extract suspected blood vessel features in the first process image through the blood vessel extraction model, separate the suspected blood vessel region from the background region, and obtain the second process image.

[0102] Specifically, the blood vessel extraction model is as follows:

[0103]

[0104] Among them, F θ σ is a two-dimensional Gaussian kernel function, where L represents the length of the segmented blood vessel and σ represents the blood vessel scale.

[0105] Since the direction of the selected blood vessel segments varies depending on the vessel's orientation, the Gaussian kernel function needs to be adjusted. This is achieved by rotating the kernel function to obtain templates with different orientations. Let A... i Let be the i-th rotation matrix that follows the deflection of the blood vessel, used to adjust the position of the pixel, and we have:

[0106]

[0107] Let f(x,y) be a point on the kernel function. After matrix operations, we obtain... but Given the rotated coordinates, the i-th matching function is:

[0108]

[0109] Where Z is the neighborhood of the template, and Z = {(u,v)||u|≤3δ,|v|≤L / 2}.

[0110] The second process image is as follows: Figure 4 As shown.

[0111] In one embodiment of the present invention, the removal module 30 is specifically used to: determine a point in the pore space with a gray value of 0, and draw a continuously expanding circle outward from the point as the center until the circle touches a point in the pore space with a gray value less than a preset threshold, and then stop expanding; store the obtained circle in an array and arrange it in ascending order of radius; sequentially take out a circle from the array and delete the circles in the array that intersect with or are contained by the circle, until the last element; set the gray value of the area where each remaining circle in the array is located to 0.

[0112] In fact, the initial purpose of the capillary bundle model is to transform irregular spatial shapes into a series of regular capillaries without affecting the particles (proliferating blood vessels) in the image. In this invention, the image obtained by matched filtering is first normalized, and the cutoff condition for circumferential expansion is reset according to the gray values. In order to retain more vascular information, 10% of the non-zero gray values ​​is used as the termination condition for circumferential expansion, that is, the preset threshold is the mean of 10% of non-zero gray values. According to the capillary bundle model, a portion of proliferating blood vessels (especially dense areas) can be largely ablated without interfering with the characteristics of the main blood vessels.

[0113] The third process image is as follows: Figure 8 As shown.

[0114] The post-processing module 40 performs threshold truncation on the third-process image by selecting an appropriate threshold based on the mean, which can further reduce interference from irrelevant features. Using a square structuring element with a parameter set to 2, dilation and erosion operations are performed on the threshold-truncated image to remove noise and pseudo-branches. The resulting final blood vessel image is shown below. Figure 9 As shown.

[0115] The method and apparatus for extracting blood vessels from OCT fundus images according to embodiments of the present invention can effectively remove interference such as vascular lesions and accurately extract clear vascular features by sequentially performing preprocessing, global feature extraction, pore filling processing based on capillary bundle model, threshold truncation processing and morphological processing on OCT fundus images, thereby greatly improving the extraction effect of blood vessels in OCT fundus images.

[0116] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. "A plurality of" means two or more, unless otherwise explicitly specified.

[0117] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0118] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0119] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0120] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0121] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0122] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0123] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0124] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0125] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for extracting blood vessels from OCT fundus images, characterized in that, Includes the following steps: The OCT fundus image is preprocessed to obtain a first process image, wherein the preprocessing includes automatic localization and removal of lesion areas; The first process image is subjected to global feature extraction by matched filtering to separate the suspected blood vessel region and obtain the second process image; The third process image is obtained by removing the interfering features in the second process image using a pore-filling method based on a capillary bundle model. The third process image is subjected to threshold truncation and morphological processing to obtain the final blood vessel image. The preprocessing of the OCT fundus image specifically includes: performing median filtering on the OCT fundus image; sharpening the median-filtered image using the Laplacian operator; automatically locating lesion regions in the sharpened image through grayscale calculation; and removing the lesion regions by overlay processing. Global feature extraction of the first process image is performed using matched filtering, specifically including: constructing a blood vessel extraction model based on a two-dimensional Gaussian kernel function to extract suspected blood vessel features in the first process image; separating the suspected blood vessel region from the background region to obtain the second process image. The method of removing interference features in the second process image by pore filling based on capillary bundle model specifically includes: determining a point with a gray value of 0 in the pore space, and drawing an expanding circle outward from the point until the circle touches a point in the pore space with a gray value less than a preset threshold, at which point the expansion stops; storing the obtained circles in an array and arranging them in ascending order of radius; sequentially taking out one circle from the array and deleting the circles that intersect with or are contained within the circle until the last element; and setting the gray value of the region containing each remaining circle in the array to 0.

2. The method for extracting blood vessels from OCT fundus images according to claim 1, characterized in that, Morphological processing of the third process image specifically includes: An expansion erosion operation is performed on the image of the third process.

3. A device for extracting blood vessels from OCT fundus images, characterized in that, include: The preprocessing module is used to preprocess the OCT fundus image to obtain a first process image, wherein the preprocessing includes automatic localization and removal of lesion areas; The extraction module is used to extract global features from the first process image through matched filtering to separate the suspected blood vessel region and obtain the second process image; The removal module is used to remove interference features in the second process image by using a pore-filling method based on a capillary bundle model to obtain the third process image; The post-processing module is used to perform threshold truncation and morphological processing on the third-process image to obtain the final blood vessel image. The preprocessing module is specifically used for: performing median filtering on the OCT fundus image; sharpening the median-filtered image using the Laplacian operator; automatically locating lesion regions in the sharpened image through grayscale calculation; and performing coverage processing on the lesion regions to remove them. The extraction module is specifically used to: construct a blood vessel extraction model based on a two-dimensional Gaussian kernel function, extract suspected blood vessel features from the first process image using the blood vessel extraction model, separate the suspected blood vessel region from the background region, and obtain the second process image. The removal module is specifically used for: determining a point in the pore space with a gray value of 0, and drawing a continuously expanding circle outward from that point until the circle encounters a point in the pore space with a gray value less than a preset threshold, at which point the expansion stops; storing the obtained circles in an array and arranging them in ascending order of radius; sequentially taking out one circle from the array and deleting circles in the array that intersect with or are contained by that circle, until the last element; and setting the gray value of the area containing each remaining circle in the array to 0.

4. The device for extracting blood vessels from OCT fundus images according to claim 3, characterized in that, The post-processing module is specifically used for: An expansion erosion operation is performed on the image of the third process.

Images