A high-definition intelligent processing method for ophthalmic OCT images
By performing longitudinal sub-block segmentation and edge detection on OCT images, the smoothness and boundary possibility of layered lines are calculated, and the pixel grayscale is adjusted, which solves the problem of edge blur in high-definition processing of OCT images, and improves the contrast and Image display effect.
Patent Information
- Application Number
- CN202411472384.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2024-12-17
- Estimated Expiration
- 2044-10-22
AI Technical Summary
The existing high-definition processing methods of OCT images lead to blurred images, especially when the image is enlarged, the edges are not clear and it is difficult to identify tiny lesions.
By performing longitudinal sub-block segmentation of ophthalmic OCT images, suspected hierarchical lines are obtained using edge detection, their smoothness and boundary possibilities are calculated, and combined with the layered gain index, the grayscale value of pixel points is adjusted to enhance image contrast.
It improves the contrast between the retinal layer in OCT images, enhances the image display effect, and helps doctors observe and identify subtle lesions more clearly.
Smart Images
Figure CN118982494B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image data processing, and in particular to a high-definition intelligent processing method for ophthalmic OCT images. Background Art
[0002] Ophthalmic OCT is a non-invasive medical imaging technology that uses low-coherence light to generate high-resolution cross-sectional images of the retina. OCT technology can accurately detect the structure of tissues at the back of the eye, helping doctors diagnose and monitor eye diseases.
[0003] In order to improve the detail resolution of the retina and other eye tissues in OCT images and assist doctors in observing and identifying subtle lesions more clearly, it is necessary to perform high-definition processing of ophthalmic OCT images, thereby improving the accuracy and reliability of diagnosis, accurately diagnosing, formulating treatment plans, and detecting potential eye diseases at an early stage.
[0004] Currently, when processing OCT images for high definition, the linear interpolation method is mainly used to improve the resolution of OCT images, but this will cause blurring of the OCT images, especially when the images are greatly magnified, the edges are no longer clear, making it difficult to identify tiny lesions in the image. Summary of the invention
[0005] The present invention provides a high-definition intelligent processing method for ophthalmic OCT images to solve the existing problems.
[0006] The present invention provides a high-definition intelligent processing method for ophthalmic OCT images using the following technical solutions:
[0007] An embodiment of the present invention provides a method for high-definition intelligent processing of ophthalmic OCT images, the method comprising the following steps:
[0008] Acquire an ophthalmic OCT image; divide the ophthalmic OCT image into a number of longitudinal sub-blocks, screen edge lines obtained by edge detection in the longitudinal sub-blocks, and obtain suspected delamination lines in the longitudinal sub-blocks;
[0009] According to the fluctuation of the distance between the pixel point on the suspected delamination line and the fitting line of the suspected delamination line, the characteristic distance of the pixel point on the suspected delamination line is obtained; according to the characteristic distance of the pixel point on the suspected delamination line, the smoothness of the suspected delamination line is obtained; according to the approximation of the extension direction of the fitting lines of different suspected delamination lines in the same longitudinal sub-block and the smoothness of the suspected delamination line, the boundary possibility of the suspected delamination line is obtained;
[0010] According to the difference in the distance between the suspected delamination lines and the demarcation possibility, the delamination gain index of the suspected delamination lines is obtained;
[0011] According to the demarcation possibility of the suspected delamination line in each longitudinal sub-block and the delamination gain index, the target grayscale of the pixel in each longitudinal sub-block is obtained; according to the target grayscale of the pixel in all longitudinal sub-blocks, the enhanced grayscale value of the pixel is obtained;
[0012] According to the enhanced grayscale values of all pixels, an ophthalmic OCT enhanced image is obtained.
[0013] Furthermore, the step of dividing the ophthalmic OCT image into a plurality of longitudinal sub-blocks includes the following specific steps:
[0014] A sliding window is constructed in the ophthalmic OCT image, wherein the length of the sliding window is 1 / 100 of the length of the ophthalmic OCT image. times, among which, To preset the sub-block ratio, the width of the sliding window is the same as the width of the ophthalmic OCT image. In the ophthalmic OCT image, the sliding window starts from the leftmost side and traverses column by column to the right until it reaches the rightmost side of the ophthalmic OCT image. For any window in the traversal process, the area composed of all the pixels in the window is recorded as a vertical sub-block.
[0015] Furthermore, the step of screening the edge lines obtained by edge detection in the vertical sub-block to obtain the suspected delamination lines in the vertical sub-block includes the following specific steps:
[0016] In any vertical sub-block, edge detection is used to obtain several edge lines; for any edge line that intersects with both the left and right boundaries of the vertical sub-block, the shortest connection path between the left and right boundaries of the vertical sub-block on the edge line is recorded as a suspected delamination line of the vertical sub-block.
[0017] Furthermore, the characteristic distance of the pixel point on the suspected delamination line is obtained according to the fluctuation of the distance between the pixel point on the suspected delamination line and the fitting straight line of the suspected delamination line, and the specific steps include the following:
[0018] In an ophthalmic OCT image, for any suspected delamination line, obtain the fitting straight line of the suspected delamination line; from the left boundary intersection point to the right boundary intersection point of the suspected delamination line and the longitudinal sub-block where it is located, sequentially obtain the distance between each pixel point on the suspected delamination line and the fitting straight line of the suspected delamination line, and record the sequence composed of the distances between all pixel points on the suspected delamination line and the fitting straight line of the suspected delamination line as the matching distance sequence of the suspected delamination line; obtain all extreme values of the matching distance sequence of the suspected delamination line; for any pixel point of the suspected delamination line, record the maximum value of the pixel point in the adjacent extreme values of the corresponding element in the matching distance sequence as the characteristic maximum value of the pixel point;
[0019] For any pixel point of the suspected layered line, the variance of all maximum values of the matching distance sequence of the suspected layered line is recorded as the first variance, and the variance of all maximum values of the matching distance sequence of the suspected layered line after excluding the characteristic maximum value of the pixel point is recorded as the second variance. The maximum value between the difference between the first variance and the second variance and 0 is taken, and the product of the maximum value and the characteristic maximum value of the pixel point is recorded as the characteristic distance of the pixel point.
[0020] Furthermore, the step of obtaining the smoothness of the suspected delamination line according to the characteristic distance of the pixel points on the suspected delamination line includes the following specific steps:
[0021] For any suspected delamination line, the mean of the characteristic distances of all pixels on the suspected delamination line is calculated, and the inverse proportional normalization result of the mean is recorded as the smoothness of the suspected delamination line.
[0022] Furthermore, the possibility of boundary of the suspected delamination line is obtained according to the approximation of the extension direction of the fitting straight lines of different suspected delamination lines in the same longitudinal sub-block and the smoothness of the suspected delamination line, and the specific steps include the following:
[0023] For any suspected delamination line, the clockwise angle between the fitting straight line of the suspected delamination line and the horizontal direction is recorded as the extension angle of the suspected delamination line;
[0024] No. The first vertical sub-block The demarcation probability of the suspected delamination line is calculated as:
[0025]
[0026] In the formula, For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The smoothness of the suspected delamination line; For the The first vertical sub-block The extension angle of the suspected delamination line; For the The vertical sub-blocks except The first line outside the suspected delamination line The extension angle of the suspected delamination line; is the cosine trigonometric function; is the absolute value function.
[0027] Furthermore, the step of obtaining the stratification gain index of the suspected stratification line according to the difference in the distance between the suspected stratification lines and the boundary possibility includes the following specific steps:
[0028] In any vertical sub-block, for any two suspected delamination lines, starting from the first column on the leftmost side of the vertical sub-block to the right, the shortest distance between the two suspected delamination lines in each column is obtained in turn, and the average value of the shortest distance between the two suspected delamination lines in all columns is recorded as the overall relative distance of the two suspected delamination lines;
[0029] No. The first vertical sub-block The calculation method of the delamination gain index of the suspected delamination line is:
[0030]
[0031] In the formula, For the The first vertical sub-block The delamination gain index of the suspected delamination line; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The possibility of demarcation of suspected delamination lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The maximum demarcation possibility of all suspected stratification lines within the area interval of the suspected stratification lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The overall relative distance of the suspected delamination lines; is the maximum value function; is a linear normalization function.
[0032] Furthermore, the step of obtaining the target grayscale of the pixel point in each vertical sub-block according to the boundary possibility and the layering gain index of the suspected layering line in each vertical sub-block includes the following specific steps:
[0033] According to the demarcation possibility and demarcation gain index of the suspected delamination line in each longitudinal sub-block, the demarcation enhancement degree of the suspected delamination line in each longitudinal sub-block is obtained;
[0034] In any vertical sub-block, all the suspected stratification lines divide the vertical sub-block into several hierarchical areas; starting from the first hierarchical area at the top, the average grayscale value of all pixels in the first hierarchical area is recorded as the characteristic grayscale of the hierarchical area, and the characteristic grayscale of each hierarchical area is calculated in sequence starting from the second hierarchical area:
[0035] Jordi In the vertical sub-block The characteristic grayscale of the level area is greater than or equal to 128, In the vertical sub-block The calculation method of the characteristic grayscale of each level area is:
[0036]
[0037] In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the stratified areas;
[0038] Jordi In the vertical sub-block The characteristic grayscale of the level area is less than 128. In the vertical sub-block The calculation method of the characteristic grayscale of each level area is:
[0039]
[0040] In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the stratified areas;
[0041] For any pixel point in any vertical sub-block, the characteristic grayscale of the hierarchical region where the pixel point is located is recorded as the target grayscale of the pixel point in the vertical sub-block.
[0042] Furthermore, the step of obtaining the boundary enhancement degree of the suspected delamination line in each longitudinal sub-block according to the boundary possibility and the delamination gain index of the suspected delamination line in each longitudinal sub-block includes the following specific steps:
[0043] No. The first vertical sub-block The calculation method of the degree of enhancement of the boundary of the suspected delamination line is:
[0044]
[0045] In the formula, For the The first vertical sub-block The degree of enhancement of the boundary between the suspected delamination lines; For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The first vertical sub-block The delamination gain index of the suspected delamination line.
[0046] Furthermore, the step of obtaining the enhanced grayscale value of the pixel point according to the target grayscale of the pixel point in all the vertical sub-blocks includes the following specific steps:
[0047] Any pixel in the ophthalmic OCT image is recorded as the target pixel; the average value of the boundary possibilities of all suspected stratification lines of the target pixel in any longitudinal sub-block where the target pixel is located is obtained, and the weighted normalization result of the average value is obtained, and the product of the weighted normalization result and the target grayscale of the target pixel in the longitudinal sub-block is recorded as the enhancement contribution index of the target pixel in the longitudinal sub-block; the sum of the enhancement contribution indices of the target pixel in all longitudinal sub-blocks where the target pixel is located is recorded as the enhancement grayscale value of the target pixel.
[0048] The beneficial effects of the technical solution of the present invention are as follows: in the process of enhancing ophthalmic OCT images, for the problem of the possibility that the suspected delamination line is the boundary line between the retinal layers of the ophthalmic OCT images, the present invention obtains the boundary possibility of the suspected delamination line through the smoothness of the suspected delamination line and the approximation of the extension direction, and distinguishes the layers of the retina of the ophthalmic OCT images; for the problem that the distance between the demarcation lines of the retina is small and the contrast between the tissues adjacent to the demarcation line may not be obvious enough, the present invention obtains the demarcation gain index of the suspected delamination line through the distance of the suspected delamination line and the boundary possibility of the suspected delamination line, and performs gain adjustment on the enhancement degree of the layer area adjacent to the demarcation line with a small distance; for the problem of high-definition processing of ophthalmic OCT images, the present invention obtains the target grayscale of the pixel point in each longitudinal sub-block through the boundary possibility of the suspected delamination line and the demarcation gain index, and then obtains the enhanced grayscale value of the pixel point in combination with the difference in the target grayscale of the pixel point in all longitudinal sub-blocks. So far, the present invention enhances the ophthalmic OCT images through the enhanced grayscale value of the pixel point, increases the contrast between the retinal layers in the ophthalmic OCT images, and improves the display effect of the ophthalmic OCT images. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0050] Figure 1 A flowchart of the steps of a high-definition intelligent processing method for ophthalmic OCT images according to the present invention;
[0051] Figure 2 This is an ophthalmic OCT image provided by the present invention. DETAILED DESCRIPTION
[0052] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the high-definition intelligent processing method of ophthalmic OCT images proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0053] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0054] The specific scheme of the high-definition intelligent processing method for ophthalmic OCT images provided by the present invention is described in detail below with reference to the accompanying drawings.
[0055] See also Figure 1 , which shows a flowchart of a method for high-definition intelligent processing of ophthalmic OCT images provided by an embodiment of the present invention, the method comprising the following steps:
[0056] Step S001: Acquire an ophthalmic OCT image; divide the ophthalmic OCT image into a number of longitudinal sub-blocks, screen the edge lines obtained by edge detection in the longitudinal sub-blocks, and obtain suspected delamination lines in the longitudinal sub-blocks.
[0057] It should be noted that ophthalmic OCT images mainly show the tissue structure at the back of the eye, especially the layers and morphology of the retina. The retina is composed of multiple layers of complex tissue. The thickness and morphological changes of each layer may indicate different types of ophthalmic diseases. The contrast of these layers helps doctors observe subtle abnormal changes more clearly and make more accurate diagnoses, so it is necessary to improve the contrast of multiple layers of tissue in ophthalmic OCT images.
[0058] It should be further explained that this embodiment mainly judges the possibility that the acquired suspected stratification line is the boundary line based on the smoothness and extension consistency of the boundary line between adjacent layers in the ophthalmic OCT image, and then enhances the contrast between layers of the ophthalmic OCT image, so the ophthalmic OCT image is first acquired.
[0059] Specifically, before acquiring OCT images, technicians calibrate the OCT device to ensure that the laser source, scanning head, and detector are in optimal condition; the patient sits in front of the device, rests his chin on a support table, and fixes his head to reduce movement;
[0060] The technician aims the OCT device’s scanning head at the patient’s eye and adjusts the device’s optical focus to precisely align it with the eye’s retina. The technician monitors the real-time preview image on the screen to ensure that the retinal layer structure is clearly visible.
[0061] The initial OCT image is acquired using an OCT device.
[0062] It should be noted that when performing high-definition processing on OCT images, in order to improve the image detail effect, the OCT images need to be enlarged first, and then the contrast between the layers of the retina needs to be enhanced.
[0063] Specifically, the length and width of the initial OCT image are enlarged by using the bilinear interpolation method. times, to obtain an ophthalmic OCT image, the ophthalmic OCT image is as follows Figure 2 As shown; among them, The magnification factor of the preset image size. This embodiment is described by taking this as an example; wherein, bilinear interpolation is a well-known technology, and the specific method is not introduced here.
[0064] Furthermore, in the ophthalmic OCT image, the gradient of each pixel is obtained by using the Sobel operator; a sliding window is constructed in the ophthalmic OCT image, and the length of the sliding window is the length of the ophthalmic OCT image. times, among which, To preset the sub-block ratio, This embodiment is described by taking this as an example. The width of the sliding window is the same as the width of the ophthalmic OCT image. In the ophthalmic OCT image, the sliding window starts from the leftmost side and traverses column by column to the right until it reaches the rightmost side of the ophthalmic OCT image. For any window in the traversal process, the area formed by all the pixels in the window is recorded as a vertical sub-block.
[0065] It should be noted that the retina is composed of multiple layers of complex tissues. There are differences in grayscale between adjacent layers of tissue, that is, the gradient of the dividing line between adjacent layers of tissue is large, and because each layer of the retinal tissue structure extends laterally, that is, the dividing line between adjacent layers of tissue also extends in the horizontal direction, the dividing line between adjacent layers of tissue will intersect with the left and right boundaries of the longitudinal sub-block, so the suspected stratification line is obtained in this way.
[0066] In any vertical sub-block, several edge lines are obtained by using Canny edge detection. For any edge line that intersects with the left and right boundaries of the vertical sub-block, the shortest connection path between the left and right boundaries of the vertical sub-block on the edge line is recorded as a suspected layering line of the vertical sub-block. Among them, Canny edge detection is a well-known technology and the specific method is not introduced here.
[0067] Step S002: According to the fluctuation of the distance between the pixel point on the suspected layering line and the fitting line of the suspected layering line, obtain the characteristic distance of the pixel point on the suspected layering line; according to the characteristic distance of the pixel point on the suspected layering line, obtain the smoothness of the suspected layering line; according to the approximation of the extension direction of the fitting lines of different suspected layering lines in the same longitudinal sub-block and the smoothness of the suspected layering line, obtain the boundary possibility of the suspected layering line.
[0068] It should be noted that when enhancing the contrast between layers of ophthalmic OCT images, it is necessary to determine the position of the boundary between layers. Among all the suspected delamination lines obtained, some of the suspected delamination lines are pseudo delamination lines formed by OCT imaging. The pseudo delamination lines exist inside the single-layer tissue area of the retina, so it is necessary to determine the boundary possibility of the suspected delamination lines. Since each layer of retinal tissue structure extends laterally, the boundary between adjacent layers of tissue is relatively smooth, and the extension direction of the boundary is similar. Therefore, the boundary possibility of the suspected delamination line is obtained through the smoothness of the suspected delamination line and the approximation of the extension direction of the fitting straight line of the suspected delamination line.
[0069] It should be further explained that when analyzing the smoothness of the suspected stratification line, due to OCT imaging, there will be burrs on the boundary between layers. The existence of burrs causes a small fluctuation in the distance between the boundary line and its fitting straight line, and the fluctuation amplitude is similar, but the pseudo-stratification line will have random larger fluctuations.
[0070] Preferably, in one embodiment of the present invention, according to the fluctuation of the distance between the pixel point on the suspected delamination line and the fitting straight line of the suspected delamination line, the characteristic distance of the pixel point on the suspected delamination line is obtained, which specifically includes:
[0071] In an ophthalmic OCT image, a rectangular coordinate system is established with the lower left corner of the ophthalmic OCT image as the origin, the horizontal direction as the horizontal axis, and the vertical direction as the vertical axis; for any suspected layering line, the least squares method is used to obtain the fitting straight line of the suspected layering line; wherein the least squares method is a well-known technology, and the specific method is not introduced here; from the left boundary intersection point to the right boundary intersection point of the suspected layering line and the longitudinal sub-block where it is located, the distance between each pixel point on the suspected layering line and the fitting straight line of the suspected layering line is obtained in sequence, and the sequence composed of the distances between all pixel points on the suspected layering line and the fitting straight line of the suspected layering line is recorded as the matching distance sequence of the suspected layering line; all extreme values of the matching distance sequence of the suspected layering line are obtained; for any pixel point of the suspected layering line, the maximum value of the pixel point in the adjacent extreme values of the corresponding element in the matching distance sequence is recorded as the characteristic maximum value of the pixel point;
[0072] For any pixel point of the suspected delamination line, the variance of all maximum values of the matching distance sequence of the suspected delamination line is recorded as the first variance, and the variance of all maximum values of the matching distance sequence of the suspected delamination line excluding the characteristic maximum value of the pixel point is recorded as the second variance. The maximum value between the difference between the first variance and the second variance and 0 is taken, and the product of the maximum value and the characteristic maximum value of the pixel point is recorded as the characteristic distance of the pixel point.
[0073] As an example, feature distance is formulated as:
[0074]
[0075] In the formula, For the The first vertical sub-block The first suspected delamination line The feature distance of pixels; For the The first vertical sub-block The variance of all maximum values of the matching distance sequence of the suspected stratification lines; For the The first vertical sub-block The first suspected delamination line After excluding the characteristic maximum value of pixels The variance of all maximum values of the matching distance sequence of the suspected stratification lines; For the The first vertical sub-block The first suspected delamination line The feature maximum value of each pixel; is the maximum value function.
[0076] It should be noted that for any pixel point on any suspected stratification line, if when calculating the variance of all maximum values of the matching distance sequence of the suspected stratification line, the variance value decreases after removing the characteristic maximum value of the pixel point, it means that the fluctuation degree of the characteristic maximum value of the pixel point is greater than the fluctuation degree of the characteristic maximum value of other pixel points, and the pixel point does not conform to the smoothness law of the boundary line of adjacent layer tissues.
[0077] It should be noted that for any suspected stratification line, the more pixels on the suspected stratification line that do not conform to the smoothness law of the boundary line between adjacent layers of tissue, that is, the more pixels on the suspected stratification line have a larger characteristic distance, the lower the smoothness of the suspected stratification line, and the less likely the suspected stratification line is to be the boundary line between retinal tissue layers.
[0078] Preferably, in one embodiment of the present invention, the smoothness of the suspected delamination line is obtained according to the characteristic distance of the pixel points on the suspected delamination line, specifically comprising: for any suspected delamination line, calculating the mean of the characteristic distances of all the pixel points on the suspected delamination line, normalizing the inverse proportion of the mean, and recording the smoothness of the suspected delamination line; this embodiment adopts Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.
[0079] It should be noted that in ophthalmic OCT images, the extension directions of the boundaries between adjacent layers of retinal tissue are similar, that is, the extension directions of the fitted straight lines of the boundaries between adjacent layers of tissue are similar. Therefore, the possibility of the boundary of the suspected stratification line is judged based on the similarity of the extension directions of the fitted straight lines of the suspected stratification lines with a higher degree of smoothness.
[0080] Preferably, in one embodiment of the present invention, the boundary possibility of the suspected delamination line is obtained according to the approximation of the extension direction of the fitting straight lines of different suspected delamination lines in the same longitudinal sub-block and the smoothness of the suspected delamination line, which specifically includes:
[0081] For any suspected delamination line, the clockwise angle between the fitting straight line of the suspected delamination line and the horizontal direction is recorded as the extension angle of the suspected delamination line;
[0082] No. The first vertical sub-block The demarcation probability of the suspected delamination line is calculated as:
[0083]
[0084] In the formula, For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The smoothness of the suspected delamination line; For the The first vertical sub-block The extension angle of the suspected delamination line; For the The vertical sub-blocks except The first line outside the suspected delamination line The extension angle of the suspected delamination line; is the cosine trigonometric function; is the absolute value function.
[0085] Step S003: Obtaining the stratification gain index of the suspected stratification line according to the difference in the distance between the suspected stratification lines and the boundary possibility.
[0086] It should be noted that in ophthalmic OCT images, the distance between the dividing lines of some parts of the retina is small. After the boundary possibility of the suspected stratification lines is obtained, when the boundary possibility is used to improve the contrast between adjacent layers of tissue in the retina, the contrast between the tissues adjacent to these boundary lines with smaller distances may not be obvious enough. Therefore, the gain adjustment is performed on the distance between the suspected stratification lines with higher boundary possibility and the degree of enhancement of the suspected stratification lines.
[0087] Preferably, in one embodiment of the present invention, the delamination gain index of the suspected delamination line is obtained according to the difference in the distance between the suspected delamination lines and the demarcation possibility, which specifically includes:
[0088] In any vertical sub-block, for any two suspected delamination lines, starting from the first column on the leftmost side of the vertical sub-block to the right, the shortest distance between the two suspected delamination lines in each column is obtained in turn, and the average value of the shortest distance between the two suspected delamination lines in all columns is recorded as the overall relative distance of the two suspected delamination lines;
[0089] No. The first vertical sub-block The calculation method of the delamination gain index of the suspected delamination line is:
[0090]
[0091] In the formula, For the The first vertical sub-block The delamination gain index of the suspected delamination line; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The possibility of demarcation of suspected delamination lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The maximum demarcation possibility of all suspected stratification lines within the area interval of the suspected stratification lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The overall relative distance of the suspected delamination lines; is the maximum value function; is a linear normalization function, and the normalization object is all the suspected delamination lines in all longitudinal sub-blocks. .
[0092] It should be noted that, in the area between any two suspected layering lines, if the boundary possibility of the suspected layering lines is relatively low, the two suspected layering lines are more likely to be adjacent boundary lines. If the overall relative distance between the two suspected layering lines is shorter, the pixel points in the adjacent areas of the two suspected layering lines need a higher degree of gain when enhancing.
[0093] Step S004: According to the boundary possibility of the suspected stratification line in each vertical sub-block and the stratification gain index, the target grayscale of the pixel point in each vertical sub-block is obtained; according to the target grayscale of the pixel point in all vertical sub-blocks, the enhanced grayscale value of the pixel point is obtained.
[0094] It should be noted that after obtaining the boundary possibility and layer gain index of the suspected layered line in the longitudinal sub-block, the ophthalmic OCT image needs to be enhanced to improve the contrast of the multi-layered tissue in the ophthalmic OCT image. First, the boundary enhancement degree of the suspected layered line is determined according to the boundary possibility and layer gain index of the suspected layered line, and then the multi-layered tissue of the ophthalmic OCT image is enhanced in turn.
[0095] It should be further explained that if the boundary possibility of the suspected stratification line is higher, the possibility that the suspected stratification line is the boundary line between adjacent layers of tissue is higher. If the stratification gain index of the suspected stratification line is higher at the same time, it means that the suspected stratification line has a higher degree of further gain in the enhancement process.
[0096] Preferably, in one embodiment of the present invention, according to the boundary possibility and the layering gain index of the suspected layering line in each vertical sub-block, the target grayscale of the pixel point in each vertical sub-block is obtained, which specifically includes:
[0097] According to the demarcation possibility and demarcation gain index of the suspected delamination line in each longitudinal sub-block, the demarcation enhancement degree of the suspected delamination line in each longitudinal sub-block is obtained;
[0098] Furthermore, in any vertical sub-block, all the suspected stratification lines divide the vertical sub-block into several hierarchical areas, and the hierarchical areas are arranged vertically; it should be noted that the pixels on the suspected stratification lines belong to the hierarchical areas on the upper side of the suspected stratification lines; starting from the first hierarchical area at the top, the average grayscale value of all pixels in the first hierarchical area is recorded as the characteristic grayscale of the hierarchical area, and then the characteristic grayscale of each hierarchical area is calculated in sequence starting from the second hierarchical area according to the following method:
[0099] Jordi In the vertical sub-block The characteristic grayscale of the level area is greater than or equal to 128, In the vertical sub-block The calculation method of the characteristic grayscale of each level area is:
[0100]
[0101] In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the different levels.
[0102] Jordi In the vertical sub-block The characteristic grayscale of the level area is less than 128. In the vertical sub-block The calculation method of the characteristic grayscale of each level area is:
[0103]
[0104] In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the different levels.
[0105] Furthermore, for any pixel point in any vertical sub-block, the characteristic grayscale of the hierarchical region where the pixel point is located is recorded as the target grayscale of the pixel point in the vertical sub-block.
[0106] It should be noted that, in any vertical sub-block, for any hierarchical area after the second hierarchical area, if the characteristic grayscale of the previous hierarchical area of the hierarchical area is closer to 255 relative to 0, the characteristic grayscale of the hierarchical area should be adjusted to 0, so as to improve the contrast between the hierarchical area and the previous hierarchical area; if the characteristic grayscale of the previous hierarchical area of the hierarchical area is closer to 0 relative to 255, the characteristic grayscale of the hierarchical area should be adjusted to 255, so as to improve the contrast between the hierarchical area and the previous hierarchical area.
[0107] Specifically, the method for obtaining the degree of boundary enhancement includes:
[0108] No. The first vertical sub-block The calculation method of the degree of enhancement of the boundary of the suspected delamination line is:
[0109]
[0110] In the formula, For the The first vertical sub-block The degree of enhancement of the boundary between the suspected delamination lines; For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The first vertical sub-block The delamination gain index of the suspected delamination line.
[0111] It should be further explained that, for any pixel in an ophthalmic OCT image, the pixel has different target grayscales in different longitudinal sub-blocks, and different longitudinal sub-blocks have different reference degrees for the pixel. The reference degree mainly depends on the clarity of the stratification between layers in the longitudinal sub-block where the pixel is located. At the fovea position, the boundaries between layers are relatively clustered, and the clarity of the stratification between layers is not obvious, which will lead to a decrease in the smoothness and similarity of the extension angles of the suspected stratification lines in the longitudinal sub-block close to the fovea position, that is, the possibility of the boundary of the suspected stratification lines in the longitudinal sub-block is reduced. Therefore, the boundary possibility of the suspected stratification lines in the longitudinal sub-block is comprehensively considered to determine the enhanced grayscale value of the pixel.
[0112] Preferably, in an embodiment of the present invention, obtaining the enhanced grayscale value of the pixel point according to the target grayscale of the pixel point in all vertical sub-blocks specifically includes:
[0113] Any pixel in the ophthalmic OCT image is recorded as the target pixel, and the average value of the boundary possibilities of all suspected stratification lines of the target pixel in any longitudinal sub-block where the target pixel is located is obtained, and the weighted normalization result of the average value is obtained. The product of the weighted normalization result and the target grayscale of the target pixel in the longitudinal sub-block is recorded as the enhancement contribution index of the target pixel in the longitudinal sub-block; the sum of the enhancement contribution indices of the target pixel in all longitudinal sub-blocks where the target pixel is located is recorded as the enhancement grayscale value of the target pixel.
[0114] As an example, the enhanced gray value is formulated as:
[0115]
[0116] In the formula, is the enhanced gray value of the target pixel; is the number of vertical sub-blocks where the target pixel is located; is the first pixel where the target pixel is located. The average of the demarcation probabilities of all suspected delamination lines in the longitudinal sub-blocks; is the first pixel where the target pixel is located. The target grayscale of the target pixel in each vertical sub-block; is the weight normalization function, and the normalization object is the weight of all vertical sub-blocks where the target pixel is located. .
[0117] Step S005: Obtain an ophthalmic OCT enhanced image according to the enhanced grayscale values of all pixels.
[0118] It should be noted that after obtaining the enhanced grayscale value of each pixel in the ophthalmic OCT image, the ophthalmic OCT image is enhanced using the enhanced grayscale value of the pixel.
[0119] Specifically, the grayscale value of each pixel in the ophthalmic OCT image is replaced with the enhanced grayscale value corresponding to each pixel to obtain an ophthalmic OCT enhanced image; the ophthalmic OCT enhanced image is the high-definition intelligent processing result of the ophthalmic OCT image.
[0120] So far, the present invention is completed.
[0121] 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, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A high-definition intelligent processing method for ophthalmic OCT images, characterized in that: The method comprises the following steps: Acquire an ophthalmic OCT image; divide the ophthalmic OCT image into a number of longitudinal sub-blocks, screen edge lines obtained by edge detection in the longitudinal sub-blocks, and obtain suspected delamination lines in the longitudinal sub-blocks; According to the fluctuation of the distance between the pixel point on the suspected delamination line and the fitting line of the suspected delamination line, the characteristic distance of the pixel point on the suspected delamination line is obtained; according to the characteristic distance of the pixel point on the suspected delamination line, the smoothness of the suspected delamination line is obtained; according to the approximation of the extension direction of the fitting lines of different suspected delamination lines in the same longitudinal sub-block and the smoothness of the suspected delamination line, the boundary possibility of the suspected delamination line is obtained; According to the difference in the distance between the suspected delamination lines and the demarcation possibility, the delamination gain index of the suspected delamination lines is obtained; According to the demarcation possibility of the suspected delamination line in each longitudinal sub-block and the delamination gain index, the target grayscale of the pixel in each longitudinal sub-block is obtained; according to the target grayscale of the pixel in all longitudinal sub-blocks, the enhanced grayscale value of the pixel is obtained; According to the enhanced gray values of all pixels, an ophthalmic OCT enhanced image is obtained.
2. According to claim 1, a high-definition intelligent processing method for ophthalmic OCT images is characterized in that: The step of dividing the ophthalmic OCT image into a plurality of longitudinal sub-blocks includes the following specific steps: A sliding window is constructed in the ophthalmic OCT image, wherein the length of the sliding window is 1 / 100 of the length of the ophthalmic OCT image. times, among which, To preset the sub-block ratio, the width of the sliding window is the same as the width of the ophthalmic OCT image. In the ophthalmic OCT image, the sliding window starts from the leftmost side and traverses column by column to the right until it reaches the rightmost side of the ophthalmic OCT image. For any window in the traversal process, the area composed of all the pixels in the window is recorded as a vertical sub-block.
3. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The step of screening edge lines obtained by edge detection in the vertical sub-block to obtain suspected delamination lines in the vertical sub-block includes the following specific steps: In any vertical sub-block, edge detection is used to obtain several edge lines; for any edge line that intersects with both the left and right boundaries of the vertical sub-block, the shortest connection path between the left and right boundaries of the vertical sub-block on the edge line is recorded as a suspected delamination line of the vertical sub-block.
4. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The specific steps of obtaining the characteristic distance of the pixel point on the suspected delamination line according to the fluctuation of the distance between the pixel point on the suspected delamination line and the fitting straight line of the suspected delamination line are as follows: In an ophthalmic OCT image, for any suspected delamination line, obtain the fitting straight line of the suspected delamination line; from the left boundary intersection point to the right boundary intersection point of the suspected delamination line and the longitudinal sub-block where it is located, sequentially obtain the distance between each pixel point on the suspected delamination line and the fitting straight line of the suspected delamination line, and record the sequence composed of the distances between all pixel points on the suspected delamination line and the fitting straight line of the suspected delamination line as the matching distance sequence of the suspected delamination line; obtain all extreme values of the matching distance sequence of the suspected delamination line; for any pixel point of the suspected delamination line, record the maximum value of the pixel point in the adjacent extreme values of the corresponding element in the matching distance sequence as the characteristic maximum value of the pixel point; For any pixel point of the suspected layered line, the variance of all maximum values of the matching distance sequence of the suspected layered line is recorded as the first variance, and the variance of all maximum values of the matching distance sequence of the suspected layered line after excluding the characteristic maximum value of the pixel point is recorded as the second variance. The maximum value between the difference between the first variance and the second variance and 0 is taken, and the product of the maximum value and the characteristic maximum value of the pixel point is recorded as the characteristic distance of the pixel point.
5. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The step of obtaining the smoothness of the suspected delamination line according to the characteristic distance of the pixel points on the suspected delamination line includes the following specific steps: For any suspected delamination line, the mean of the characteristic distances of all pixels on the suspected delamination line is calculated, and the inverse proportional normalization result of the mean is recorded as the smoothness of the suspected delamination line.
6. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The method of obtaining the boundary possibility of the suspected delamination line according to the approximation of the extension direction of the fitting straight lines of different suspected delamination lines in the same longitudinal sub-block and the smoothness of the suspected delamination line includes the following specific steps: For any suspected delamination line, the clockwise angle between the fitting straight line of the suspected delamination line and the horizontal direction is recorded as the extension angle of the suspected delamination line; No. The first vertical sub-block The demarcation probability of the suspected delamination line is calculated as: In the formula, For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The smoothness of the suspected delamination line; For the The first vertical sub-block The extension angle of the suspected delamination line; For the The vertical sub-blocks except The first line outside the suspected delamination line The extension angle of the suspected delamination line; is the cosine trigonometric function; is the absolute value function.
7. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The step of obtaining the stratification gain index of the suspected stratification line according to the difference in the distance between the suspected stratification lines and the demarcation possibility includes the following specific steps: In any vertical sub-block, for any two suspected delamination lines, starting from the first column on the leftmost side of the vertical sub-block to the right, the shortest distance between the two suspected delamination lines in each column is obtained in turn, and the average value of the shortest distance between the two suspected delamination lines in all columns is recorded as the overall relative distance of the two suspected delamination lines; No. The first vertical sub-block The calculation method of the delamination gain index of the suspected delamination line is: In the formula, For the The first vertical sub-block The delamination gain index of the suspected delamination line; For the The number of suspected delamination lines in each longitudinal sub-block; For the The vertical sub-blocks except The first line outside the suspected delamination line The possibility of demarcation of suspected delamination lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The maximum demarcation possibility of all suspected stratification lines within the area interval of the suspected stratification lines; For the In the vertical sub-block Suspected delamination lines to except the first The first line outside the suspected delamination line The overall relative distance of the suspected delamination lines; is the maximum value function; is a linear normalization function.
8. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The specific steps of obtaining the target grayscale of the pixel point in each vertical sub-block according to the boundary possibility and the layering gain index of the suspected layering line in each vertical sub-block are as follows: According to the demarcation possibility and demarcation gain index of the suspected delamination line in each longitudinal sub-block, the demarcation enhancement degree of the suspected delamination line in each longitudinal sub-block is obtained; In any vertical sub-block, all the suspected stratification lines divide the vertical sub-block into several hierarchical areas; starting from the top first hierarchical area, the average grayscale value of all pixels in the first hierarchical area is recorded as the characteristic grayscale of the hierarchical area, and the characteristic grayscale of each hierarchical area is calculated in sequence starting from the second hierarchical area: Jordi In the vertical sub-block The characteristic grayscale of the level area is greater than or equal to 128, In the vertical sub-block The calculation method of the characteristic grayscale of each level area is: In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the stratified areas; Jordi In the vertical sub-block The characteristic grayscale of the level area is less than 128. In the vertical sub-block The calculation method of the characteristic grayscale of each level area is: In the formula, For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The characteristic grayscale of each level area; For the In the vertical sub-block The level area and The degree of enhancement of the boundary between the suspected stratification lines of the stratified areas; For any pixel point in any vertical sub-block, the characteristic grayscale of the hierarchical region where the pixel point is located is recorded as the target grayscale of the pixel point in the vertical sub-block.
9. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 8, characterized in that: The step of obtaining the boundary enhancement degree of the suspected delamination line in each longitudinal sub-block according to the boundary possibility and the delamination gain index of the suspected delamination line in each longitudinal sub-block includes the following specific steps: No. The first vertical sub-block The calculation method of the degree of enhancement of the boundary of the suspected delamination line is: In the formula, For the The first vertical sub-block The degree of enhancement of the boundary between the suspected delamination lines; For the The first vertical sub-block The possibility of demarcation of suspected delamination lines; For the The first vertical sub-block The delamination gain index of the suspected delamination line.
10. The method for high-definition intelligent processing of ophthalmic OCT images according to claim 1, characterized in that: The step of obtaining the enhanced grayscale value of the pixel point according to the target grayscale of the pixel point in all the vertical sub-blocks includes the following specific steps: Any pixel in the ophthalmic OCT image is recorded as the target pixel; the average value of the boundary possibilities of all suspected stratification lines of the target pixel in any longitudinal sub-block where the target pixel is located is obtained, and the weighted normalization result of the average value is obtained, and the product of the weighted normalization result and the target grayscale of the target pixel in the longitudinal sub-block is recorded as the enhancement contribution index of the target pixel in the longitudinal sub-block; the sum of the enhancement contribution indices of the target pixel in all longitudinal sub-blocks where the target pixel is located is recorded as the enhancement grayscale value of the target pixel.