Feature extraction method and registration method for vascular structure in retina OCT image
Through the feature extraction method of vascular structure in retinal OCT images, combining 3D, intersection point and SURF features, combined with pixel point importance evaluation model and vascular skeleton transformation, the CPD algorithm is used to accurately register retinal OCT images, which solves the existing technology The issues of accuracy and time complexity improve the accuracy and speed of registration.
Patent Information
- Application Number
- CN202411228335.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-03
AI Technical Summary
The prior art is difficult to take into account accuracy and time complexity in retinal OCT image registration, especially when the vascular structure is damaged or there are many features, resulting in computational complexity and noise sensitivity problems.
A feature extraction method for vascular structures in retinal OCT images is proposed. Through the fusion of 3D features, intersection features and SURF features, combined with pixel point importance evaluation model and vascular skeleton image transformation, the CPD algorithm is used for precise registration.
It realizes a more comprehensive extraction of vascular structure information, improves the accuracy and speed of registration, is suitable for clinical diagnosis, and reduces calculation costs and noise sensitivity.
Smart Images

Figure CN118968574A_ABST
Abstract
Description
[0001] Technology Neighborhood
[0002] The present invention belongs to the technical field of medical image processing technology, and in particular relates to a feature extraction method and a registration method for blood vessel structures in retinal OCT images. Background Art
[0003] Optical coherence tomography (OCT) is a non-invasive, high-resolution imaging technology widely used in ophthalmology and other medical fields. It can produce high-resolution cross-sectional images of each layer of the retina. These images can be used to diagnose and monitor a variety of eye diseases, such as macular degeneration, glaucoma, retinal detachment, retinal vascular occlusion, etc., and play a vital role in ophthalmic diagnosis. The OCT enface image is an image that projects the retinal OCT three-dimensional volume data from top to bottom along the Z-axis onto a two-dimensional plane. This projection can be used to visualize and analyze retinal images at a fixed depth or volume, allowing doctors to intuitively observe and analyze the retinal structure.
[0004] In order to achieve accurate monitoring of a variety of ophthalmic diseases, it is necessary to register and align OCT enface images at adjacent time points. This process involves spatially aligning (i.e., registering) enface images acquired at different time points through image processing technology to ensure the comparability between images. The registered images can intuitively show the changes in the lesion area, including the growth and atrophy of new blood vessels, and the improvement or deterioration of retinal structure. This is of great significance for evaluating the treatment effect and guiding the formulation of subsequent treatment strategies. However, due to eye movement, interference from the lesion itself, etc., registration has brought great difficulties, which has brought inconvenience to the subsequent observation and research of the lesion. Therefore, it is necessary to develop a technology that can achieve accurate registration of OCT enface images of adjacent time series.
[0005] There are many methods for implementing enface image registration, which can be mainly divided into two categories: feature-based methods and intensity-based methods. Feature-based registration methods extract key feature points from two images and then establish the correspondence between images by matching these feature points. Intensity-based methods directly compare the pixel intensities of the images and find the best registration parameters by optimizing an objective function (such as minimizing the mean square error between the two images). Intensity-based methods do not require feature extraction and directly calculate pixel intensity, so they are more applicable to images. However, this type of method is more sensitive to noise and has poor registration accuracy in complex images. Considering that OCT enface contains speckle noise, feature-based image registration methods are very effective in processing OCT enface images.
[0006] Among the feature-based methods, the most commonly used one is the registration method based on the feature points of the vascular structure. This method extracts the vascular structure from the retinal image, uses the vascular structure for feature description, matches the features of the two images and estimates the transformation model, and finally achieves image registration. However, this type of method is highly dependent on vascular features, and the appearance of lesions often damages and affects the vascular structure, thereby affecting the matching of vascular features. And if there are many features in the image, it will also bring computational complexity, affecting the computational efficiency and computational cost, and bringing inconvenience to clinical diagnosis. At present, no literature has proposed a method that can comprehensively consider and take into account the accuracy and time complexity of image registration.
[0007] Therefore, how to propose a feature extraction method and registration method for vascular structures in retinal OCT images to comprehensively consider and balance the accuracy and time complexity of image registration is an urgent problem to be solved by technicians in this field. Summary of the invention
[0008] In order to solve the problems existing in the prior art, the present invention proposes a feature extraction method and a registration method for vascular structures in retinal OCT images.
[0009] The technical solution of the present invention is as follows:
[0010] A method for extracting features of vascular structures in retinal OCT images, comprising:
[0011] Extracting a first feature for reflecting a 3D feature of a vascular structure from the retinal OCT image to obtain a first feature;
[0012] Converting the retinal OCT image into a vascular skeleton image to obtain a vascular skeleton image;
[0013] Performing a second feature extraction for reflecting the intersection feature of the blood vessel structure and a third feature extraction for reflecting the SURF feature of the blood vessel structure on the blood vessel skeleton image, and obtaining the second feature and the third feature accordingly;
[0014] The first feature, the second feature and the third feature are fused with respect to feature coordinates to form a combined feature as a feature extraction result.
[0015] Furthermore, the specific steps of extracting a first feature for reflecting the 3D feature of the vascular structure from the retinal OCT image to obtain the first feature include:
[0016] 1) performing denoising processing on each B-scan image corresponding to the retinal OCT image;
[0017] 2) For each B-scan image after denoising, calculate the median, mean and variance of each cylinder;
[0018] 3) constructing a pixel importance evaluation model for the retinal OCT image based on the calculated median, mean and variance, and calculating the intensity difference of cylindrical pixels in the retinal OCT image based on the pixel importance evaluation model for the retinal OCT image;
[0019] 4) Compare the intensity difference of the cylindrical pixel points with a set threshold value related to the median value, select the pixel points that are greater than or equal to the set threshold value, and use the selected pixel points as the first feature reflecting the 3D feature of the blood vessel structure.
[0020] Furthermore, the expression of the pixel importance evaluation model is:
[0021]
[0022] Where f(x) (s,i) is the importance evaluation result of the pixel point in the ith column of the sth B-scan image; M (s,i) is the pixel mean of the ith column of the sth B-scan image; σ (s,i) is the pixel variance of the ith column of the sth B-scan image; e is a mathematical constant;
[0023] The calculation formula for the intensity difference of the cylindrical pixel point is:
[0024] Sub (s,i,j) =|I (s,i,j) ×(1-f(x) (s,i) )|
[0025] Where, Sub (s,i,j) is the grayscale difference estimation function value of the pixel in the ith column and jth row of the sth B-scan image; I (s,i,j) is the pixel value of the i-th column and j-th row of the s-th B-scan image.
[0026] Furthermore, the specific steps of converting the retinal OCT image into a vascular skeleton image to obtain the vascular skeleton image include:
[0027] 1) Projecting the retinal OCT image in the Z-axis direction to obtain an OCT enface image;
[0028] 2) Extracting the vascular skeleton image based on the OCT enface image.
[0029] Furthermore, the specific steps of extracting the second feature for reflecting the intersection feature of the blood vessel structure include:
[0030] 1) Identification of vascular structure intersections: neighborhood graph analysis of vascular skeleton images to obtain potential vascular structure intersections;
[0031] 2) Extraction of vascular structure intersections: Based on the pre-constructed first feature extraction matrix, potential vascular intersections are extracted to obtain vascular intersections after preliminary screening;
[0032] Refined processing of vascular structure intersections: based on the pre-constructed second feature extraction matrix, pseudo feature points are eliminated from the vascular structure intersections after the initial screening to obtain the vascular intersections after the secondary screening;
[0033] Marking and application of vascular structure intersections,The vascular intersections after secondary screening are marked as the second feature extraction,which reflects the characteristics of vascular structure intersections.
[0034] Furthermore, the expression of the first feature extraction matrix T1 is:
[0035]
[0036] The specific extraction method of extracting vascular structure intersections from potential vascular intersections based on the pre-constructed first feature extraction matrix to obtain vascular intersections after preliminary screening is as follows: using the first feature extraction matrix to sum eight neighborhood values for each pixel in the OCT enface image, and screening out pixels whose summation results are greater than a first set value as potential vascular structure intersections, wherein the summation expression is:
[0037] g r1 (x,y)=f(x-1,y-1)+f(x-1,y)+f(x-1,y+1)+f(x,y-1)+f(x,y)
[0038] +f(x,y+1)+f(x+1,y-1)+f(x+1,y)+f(x+1,y+1)
[0039] In the formula, g r1 (x,y) is the sum of the eight neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x-1,y-1), f(x-1,y), f(x-1,y+1), f(x,y-1), f(x,y), f(x,y+1), f(x+1,y-1), f(x+1,y+1) and f(x+1,y+1) are the pixels with coordinates (x-1,y-1), (x-1,y), (x-1,y+1), (x,y-1), (x,y), (x,y+1), (x+1,y-1), (x+1,y) and (x+1,y+1) in the OCT enface image respectively.
[0040] Furthermore, the expression of the second feature extraction matrix T2 is:
[0041]
[0042] The specific method of eliminating pseudo feature points from the intersection points of the vascular structures after the primary screening based on the pre-constructed second feature extraction matrix to obtain the vascular intersection points after the secondary screening includes: using the first feature extraction matrix T2 to sum the values of the 8 neighborhoods of the vascular intersection points after the primary screening, and screening out pixel points whose summation results are greater than or equal to the second set value or equal to the third set value as potential vascular intersection points after the secondary screening; wherein the summation expression is:
[0043] g r2 (x,y)=f(x,y-1)+f(x-1,y)+f(x,y)+f(x+1,y)+f(x,y+1)
[0044] In the formula, g r2 (x,y) is the sum of the five neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x,y-1), f(x-1,y), f(x,y), f(x+1,y) and f(x,y+1) are the pixels with coordinates (x,y-1), (x-1,y), (x,y), (x+1,y) and (x,y+1) in the OCT enface image respectively.
[0045] Furthermore, the specific steps of extracting the third feature for reflecting the SURF feature of the vascular structure include:
[0046] The SURF algorithm is used to extract the third feature for reflecting the SURF feature of the vascular structure, and the extraction result is used as the third feature.
[0047] A registration method for vascular structures in retinal OCT images, comprising:
[0048] Acquire two adjacent time retinal OCT images;
[0049] The method for extracting the feature of the vascular structure in the retinal OCT image of the present invention is used to extract the combined features of the two adjacent time retinal OCT images respectively, so as to obtain two feature extraction results corresponding to the two adjacent time retinal OCT images;
[0050] The two adjacent time retinal OCT images are respectively converted into vascular skeleton images to obtain corresponding vascular skeleton images, and then the vascular skeleton features are extracted and the coordinates are simplified for each of the vascular skeleton images to obtain two vascular skeleton features corresponding to the two adjacent time retinal OCT images;
[0051] Based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, the vascular structures of the two adjacent time retinal OCT images are registered.
[0052] Furthermore, the specific method for performing vascular structure registration of the two adjacent time retinal OCT images based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images includes:
[0053] Based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, the CPD algorithm is used to perform vascular structure registration of the two adjacent time retinal OCT images.
[0054] Compared with the prior art, the present invention has the following beneficial effects:
[0055] The present invention proposes a feature extraction method for vascular structure in retinal OCT images. The method extracts a first feature reflecting the 3D feature of the vascular structure, a second feature reflecting the intersection feature of the vascular structure, and a third feature reflecting the SURF feature of the vascular structure based on the retinal OCT image, and fuses the extracted features. Therefore, the obtained features can have more comprehensive vascular structure information, laying a foundation for subsequent operations such as registration.
[0056] The feature extraction method of the present invention designs a pixel importance evaluation model and screens the first feature that reflects the 3D features of the vascular structure based on the intensity difference of the cylindrical pixels, which is used to supplement 2D features such as the intersection features of the vascular structure and the SURF features, and extracts the vascular structure features in multiple dimensions.
[0057] The feature extraction method of the present invention designs an extraction step of a second feature extraction for reflecting the characteristics of the intersection of the vascular structure. The potential vascular intersections are extracted by a first feature extraction matrix to obtain the vascular intersections after the initial screening. Then, pseudo feature points are eliminated from the vascular intersections after the initial screening by a second feature extraction matrix to obtain the vascular intersections after the secondary screening, thereby avoiding the omission of vascular intersections and making the extraction of the second feature more accurate.
[0058] The feature extraction method of the present invention adopts Douglas-Peucker to simplify the coordinates of the blood vessel points, which meets the requirement of rapid feature extraction and is particularly suitable for operations such as clinical registration.
[0059] The present invention provides a registration method for vascular structures in retinal OCT images. The method adopts the feature extraction method of the present invention and can optimize the accuracy while ensuring the registration speed. BRIEF DESCRIPTION OF THE DRAWINGS
[0060] Figure 1Schematic diagram of the process of the feature extraction method of the vascular structure in the retinal OCT image in the embodiment;
[0061] Figure 2 A schematic diagram of a flow chart of a registration method for vascular structures in retinal OCT images in an embodiment;
[0062] FIG3( a ) is one of the projection diagrams in the embodiment;
[0063] FIG3( b ) is one of the projection diagrams in the embodiment;
[0064] Figure 4 Schematic diagram of 3D feature points in the embodiment;
[0065] Figure 5 Schematic diagram of blood vessel intersection identification in an embodiment;
[0066] Figure 6 Schematic diagram of pseudo feature points in the embodiment;
[0067] Figure 7 This is a schematic diagram of blood vessel intersection marking in the embodiment;
[0068] Figure 8 This is a simplified schematic diagram of blood vessel coordinates in the embodiment;
[0069] Fig. 9 Schematic diagram of image registration results in the embodiment. DETAILED DESCRIPTION
[0070] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent modifications of the present invention by technicians in this field all fall within the scope defined by the claims attached to this application.
[0071] Embodiment 1:
[0072] The present invention provides a method for extracting features of vascular structures in retinal OCT images, such as Figure 1 and Figure 2 As shown, including:
[0073] S1, extracting a first feature for reflecting the 3D feature of the vascular structure from the retinal OCT image to obtain a first feature;
[0074] S2, converting the retinal OCT image into a vascular skeleton image to obtain a vascular skeleton image;
[0075] S3, extracting the second feature for reflecting the intersection feature of the vascular structure and extracting the third feature for reflecting the SURF feature of the vascular structure from the vascular skeleton image, and obtaining the second feature and the third feature accordingly;
[0076] S4. Fusing the first feature, the second feature, and the third feature with respect to the feature coordinates to form a combined feature as a feature extraction result.
[0077] Furthermore, the first feature for reflecting the 3D feature of the vascular structure is extracted from the retinal OCT image, and the specific steps of obtaining the first feature include:
[0078] 1) De-noising each B-scan image corresponding to the retinal OCT image, because the retinal OCT image may generate noise that is difficult to remove during the imaging process, especially multiplicative noise. Specifically, the BM3D algorithm can be used for denoising to effectively remove these noises and improve the accuracy of subsequent feature extraction;
[0079] 2) For each B-scan image after denoising, i.e., each slice, a top-down projection in the Z-axis direction is performed, as shown in FIG3(a) and FIG3(b), and the median, mean, and variance of each cylinder (Z direction) are calculated;
[0080] 3) Based on the calculated median, mean and variance, a pixel importance evaluation model of the retinal OCT image is constructed, and based on the pixel importance evaluation model of the retinal OCT image, the intensity difference of the cylindrical pixels in the retinal OCT image is calculated;
[0081] 4) Compare the intensity difference of the cylindrical pixels with the set threshold value related to the median value, and select the pixels that are greater than or equal to the set threshold value (i.e., the pixels with significant changes), such as Figure 4 As shown, the screened pixel points are used as the first feature reflecting the 3D features of the blood vessel structure, and a first feature point set can be further formed.
[0082] The threshold value related to the median can be set as: α×K (s,i) ; In the formula, α is the threshold coefficient, which is generally taken as 0.0001; K (s,i) is the median value of the pixels in the i-th column of the s-th B-scan image;
[0083] Furthermore, the expression of the pixel importance evaluation model is:
[0084]
[0085] Where f(x) (s,i) is the importance evaluation result of the pixel point (also called voxel point) in the ith column of the sth B-scan image; M (s,i) is the pixel mean of the ith column of the sth B-scan image; σ (s,i) is the pixel variance of the i-th column of the s-th B-scan image, i∈[1,n]; e is a mathematical constant;
[0086] The calculation formula for the intensity difference of cylindrical pixels is:
[0087] Sub (s,i,j) =|I (s,i,j) ×(1-f(x) (s,i) )
[0088] Where, Sub (s,i,j) is the grayscale difference estimation function value of the pixel in the ith column and jth row of the sth B-scan image; I (s,i,j) is the pixel value of the i-th column and j-th row of the s-th B-scan image;
[0089] Furthermore, the retinal OCT image is converted into a vascular skeleton image, and the specific steps of obtaining the vascular skeleton image include:
[0090] 1) Project the retinal OCT image in the Z-axis direction to obtain the OCT enface image;
[0091] 2) Extract the vascular skeleton image based on the OCT enface image;
[0092] 3) The coordinates of the vascular skeleton points in the extracted vascular skeleton image can be simplified using the Douglas-Peucker method.
[0093] Furthermore, the specific steps of extracting the second feature for reflecting the intersection feature of the blood vessel structure include:
[0094] 1) Identification of vascular structure intersections: Perform neighborhood graph analysis on the vascular skeleton image to obtain potential vascular structure intersections, such as Figure 5 As shown;
[0095] 2) Extraction of vascular structure intersections: Based on the pre-constructed first feature extraction matrix, potential vascular intersections are extracted to obtain vascular intersections after preliminary screening;
[0096] The refined processing of the intersection of vascular structures is based on the pre-constructed second feature extraction matrix to eliminate pseudo feature points of the intersection of vascular structures after the initial screening, and obtain the intersection of vascular structures after the secondary screening, such as Figure 6 As shown;
[0097] Marking and application of vascular structure intersections, after secondary screening, the vascular intersections are marked as the second feature extraction that reflects the characteristics of the vascular structure intersections, and further a second feature point set can be formed. After extracting the vascular intersections, the pixels of the entire binary image are replaced with 0 and 1. The background of the vascular image becomes white (grayscale value is 1), the vascular pixels become black (grayscale value is 0), and the extracted intersections are marked to facilitate more intuitive observation of the image, such as Figure 7 shown.
[0098] Furthermore, the expression of the first feature extraction matrix T1 is:
[0099]
[0100] Based on the pre-constructed first feature extraction matrix, the potential vascular intersections are extracted for vascular structure intersections. The specific extraction method for obtaining the vascular intersections after preliminary screening is as follows: the first feature extraction matrix is used to sum the 8 neighborhood values of each pixel in the OCT enface image, and the pixel points whose summation result is greater than the first set value (3 in this example) are screened out as potential vascular structure intersections, where the summation expression is:
[0101] g r1 (x,y)=f(x-1,y-1)+f(x-1,y)+f(x-1,y+1)+f(x,y-1)+f(x,y)
[0102] +f(x,y+1)+f(x+1,y-1)+f(x+1,y)+f(x+1,y+1)
[0103] In the formula, g r1 (x,y) is the sum of the eight neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x-1,y-1), f(x-1,y), f(x-1,y+1), f(x,y-1), f(x,y), f(x,y+1), f(x+1,y-1), f(x+1,y+1) and f(x+1,y+1) are the pixels with coordinates (x-1,y-1), (x-1,y), (x-1,y+1), (x,y-1), (x,y), (x,y+1), (x+1,y-1), (x+1,y) and (x+1,y+1) in the OCT enface image respectively.
[0104] Furthermore, the expression of the second feature extraction matrix T2 is:
[0105]
[0106] Based on the pre-constructed second feature extraction matrix, the pseudo feature points of the intersection points of the vascular structures after the primary screening are eliminated, and the specific method for obtaining the vascular intersection points after the secondary screening includes: using the first feature extraction matrix T2 to sum the values of the 8 neighborhoods of the vascular intersection points after the primary screening, and screening out the pixel points whose summation result is greater than or equal to the second set value or equal to the third set value as the potential vascular intersection points after the secondary screening; wherein the summation expression is:
[0107] g r2(x,y)=f(x,y-1)+f(x-1,y)+f(x,y)+f(x+1,y)+f(x,y+1)
[0108] In the formula, g r2 (x,y) is the sum of the five neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x,y-1), f(x-1,y), f(x,y), f(x+1,y) and f(x,y+1) are the pixels with coordinates (x,y-1), (x-1,y), (x,y), (x+1,y) and (x,y+1) in the OCT enface image respectively.
[0109] Furthermore, the specific steps of extracting the third feature for reflecting the SURF feature of the vascular structure include:
[0110] The SURF algorithm is used to extract the third feature for reflecting the SURF feature of the blood vessel structure, and the extraction result is used as the third feature, which can further form a third feature point set.
[0111] Embodiment 2:
[0112] A registration method for vascular structures in retinal OCT images of the present invention is as follows: Figure 2 As shown, including:
[0113] Acquire two adjacent time retinal OCT images;
[0114] The feature extraction method of the vascular structure in the retinal OCT image of the present invention is used to extract the combined features of two adjacent time retinal OCT images respectively, and obtain two feature extraction results corresponding to the two adjacent time retinal OCT images;
[0115] The vascular skeleton images of two adjacent retinal OCT images are transformed respectively to obtain the corresponding vascular skeleton images. The vascular skeleton features of each vascular skeleton image are extracted and the coordinates are simplified to obtain two vascular skeleton features corresponding to the two adjacent retinal OCT images. Further, a vascular skeleton feature point set can be formed. The results are as follows: Figure 8 As shown;
[0116] Based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, the vascular structure registration of the two adjacent time retinal OCT images is performed.
[0117] Furthermore, based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, a specific method for vascular structure registration of the two adjacent time retinal OCT images includes:
[0118] Based on the feature extraction results and vascular skeleton features of two adjacent time retinal OCT images, the CPD algorithm is used to perform vascular structure registration of two adjacent time retinal OCT images. The results are shown in Figure 2. Fig. 9 shown.
[0119] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for extracting features of vascular structures in retinal OCT images, characterized in that: include: Extracting a first feature for reflecting a 3D feature of a vascular structure from the retinal OCT image to obtain a first feature; Converting the retinal OCT image into a vascular skeleton image to obtain a vascular skeleton image; Performing a second feature extraction for reflecting the intersection feature of the blood vessel structure and a third feature extraction for reflecting the SURF feature of the blood vessel structure on the blood vessel skeleton image, and obtaining the second feature and the third feature accordingly; The first feature, the second feature and the third feature are fused with respect to feature coordinates to form a combined feature as a feature extraction result.
2. The method for extracting the feature of vascular structure in retinal OCT images according to claim 1, characterized in that: The specific steps of extracting the first feature for reflecting the 3D feature of the vascular structure from the retinal OCT image to obtain the first feature include: 1) performing denoising processing on each B-scan image corresponding to the retinal OCT image; 2) For each B-scan image after denoising, calculate the median, mean and variance of each cylinder; 3) constructing a pixel importance evaluation model for the retinal OCT image based on the calculated median, mean and variance, and calculating the intensity difference of cylindrical pixels in the retinal OCT image based on the pixel importance evaluation model for the retinal OCT image; 4) Compare the intensity difference of the cylindrical pixel points with a set threshold value related to the median value, select the pixel points that are greater than or equal to the set threshold value, and use the selected pixel points as the first feature reflecting the 3D feature of the blood vessel structure.
3. The method for extracting features of vascular structures in retinal OCT images according to claim 2, characterized in that: The expression of the pixel importance evaluation model is: Where f(x) (s,i) is the importance evaluation result of the pixel point in the ith column of the sth B-scan image; M (s,i) is the pixel mean of the ith column of the sth B-scan image; σ (s,i) is the pixel variance of the ith column of the sth B-scan image; e is a mathematical constant; The calculation formula for the intensity difference of the cylindrical pixel point is: Sub (s,i,j) =|I (s,i,j) ×(1-f(x) (s,i) )| Where, Sub (s,i,j) is the grayscale difference estimation function value of the pixel in the ith column and jth row of the sth B-scan image; I (s,i,j) is the pixel value of the i-th column and j-th row of the s-th B-scan image.
4. The method for extracting features of vascular structures in retinal OCT images according to claim 1, characterized in that: The specific steps of converting the retinal OCT image into a vascular skeleton image to obtain the vascular skeleton image include: 1) Projecting the retinal OCT image in the Z-axis direction to obtain an OCT enface image; 2) Extracting the vascular skeleton image based on the OCT enface image.
5. The method for extracting features of vascular structures in retinal OCT images according to claim 4, characterized in that: The specific steps of extracting the second feature for reflecting the intersection feature of the vascular structure include: 1) Identification of vascular structure intersections: neighborhood graph analysis of vascular skeleton images to obtain potential vascular structure intersections; 2) Extraction of vascular structure intersections: Based on the pre-constructed first feature extraction matrix, potential vascular intersections are extracted to obtain vascular intersections after preliminary screening; Refined processing of vascular structure intersections: based on the pre-constructed second feature extraction matrix, pseudo feature points are eliminated from the vascular structure intersections after the initial screening to obtain the vascular intersections after the secondary screening; Marking and application of vascular structure intersections,The vascular intersections after secondary screening are marked as the second feature extraction,which reflects the characteristics of vascular structure intersections.
6. The method for extracting features of vascular structures in retinal OCT images according to claim 5, characterized in that: The expression of the first feature extraction matrix T1 is: The specific extraction method of extracting vascular structure intersections from potential vascular intersections based on the pre-constructed first feature extraction matrix to obtain vascular intersections after preliminary screening is as follows: using the first feature extraction matrix to sum eight neighborhood values for each pixel in the OCT enface image, and screening out pixels whose summation results are greater than a first set value as potential vascular structure intersections, wherein the summation expression is: g r1 (x,y)=f(x-1,y-1)+f(x-1,y)+f(x-1,y+1)+f(x,y-1)+f(x,y) +f(x,y+1)+f(x+1,y-1)+f(x+1,y)+f(x+1,y+1) In the formula, g r1 (x,y) is the sum of the eight neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x-1,y-1), f(x-1,y), f(x-1,y+1), f(x,y-1), f(x,y), f(x,y+1), f(x+1,y-1), f(x+1,y+1) and f(x+1,y+1) are the pixels with coordinates (x-1,y-1), (x-1,y), (x-1,y+1), (x,y-1), (x,y), (x,y+1), (x+1,y-1), (x+1,y) and (x+1,y+1) in the OCT enface image respectively.
7. The method for extracting features of vascular structures in retinal OCT images according to claim 5, characterized in that: The expression of the second feature extraction matrix T2 is: The specific method of eliminating pseudo feature points from the intersection points of the vascular structures after the primary screening based on the pre-constructed second feature extraction matrix to obtain the vascular intersection points after the secondary screening includes: using the first feature extraction matrix T2 to sum the values of the five neighborhoods of the vascular intersection points after the primary screening, and screening out pixel points whose summation results are greater than or equal to the second set value or equal to the third set value as potential vascular intersection points after the secondary screening, wherein the summation expression is: g r2 (x,y)=f(x,y-1)+f(x-1,y)+f(x,y)+f(x+1,y)+f(x,y+1) In the formula, g r2 (x,y) is the sum of the five neighborhood values corresponding to the pixel with coordinates (x,y) in the OCT enface image; f(x,y-1), f(x-1,y), f(x,y), f(x+1,y) and f(x,y+1) are the pixels with coordinates (x,y-1), (x-1,y), (x,y), (x+1,y) and (x,y+1) in the OCT enface image respectively.
8. The method for extracting features of vascular structures in retinal OCT images according to claim 1, characterized in that: The specific steps of extracting the third feature for reflecting the SURF feature of the vascular structure include: The SURF algorithm is used to extract the third feature for reflecting the SURF feature of the vascular structure, and the extraction result is used as the third feature.
9. A method for registering vascular structures in retinal OCT images, characterized in that: include: Acquire two adjacent time retinal OCT images; Using the feature extraction method of vascular structure in retinal OCT images according to any one of claims 1 to 7, respectively extracting combined features of the two adjacent time retinal OCT images to obtain two feature extraction results corresponding to the two adjacent time retinal OCT images; The two adjacent time retinal OCT images are respectively converted into vascular skeleton images to obtain corresponding vascular skeleton images, and then the vascular skeleton features are extracted and the coordinates are simplified for each of the vascular skeleton images to obtain two vascular skeleton features corresponding to the two adjacent time retinal OCT images; Based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, the vascular structures of the two adjacent time retinal OCT images are registered.
10. The registration method for vascular structures in retinal OCT images according to claim 7, characterized in that: The specific method for performing vascular structure registration of the two adjacent time retinal OCT images based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images includes: Based on the feature extraction results and vascular skeleton features corresponding to the two adjacent time retinal OCT images, the CPD algorithm is used to perform vascular structure registration of the two adjacent time retinal OCT images.
Citation Information
Patent Citations
Three-dimensional retina OCT image vascular detection and registration method, equipment and application
CN108257126A
Motion artifact correction method for time sequence three-dimensional retina SD-OCT image
CN110033496A
Retina image registration method based on expansion vessel constraint and metric learning
CN118447063A
Using multiple sub-volumes, thicknesses, and curvatures for oct / OCTA data registration and retinal landmark detection
WO2022117661A2
Cited By
Automatic identification method and system for OCT image blood vessel structure
CN121190435A
