CN114092405A - Retina layer automatic segmentation method for macular edema OCT image - Google Patents
Retina layer automatic segmentation method for macular edema OCT image Download PDFInfo
- Publication number
- CN114092405A CN114092405A CN202111241258.7A CN202111241258A CN114092405A CN 114092405 A CN114092405 A CN 114092405A CN 202111241258 A CN202111241258 A CN 202111241258A CN 114092405 A CN114092405 A CN 114092405A
- Authority
- CN
- China
- Prior art keywords
- image
- edema
- retina
- segmentation
- macular edema
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 210000001525 retina Anatomy 0.000 title claims abstract description 54
- 230000011218 segmentation Effects 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 40
- 208000001344 Macular Edema Diseases 0.000 title claims abstract description 33
- 206010025415 Macular oedema Diseases 0.000 title claims abstract description 33
- 201000010230 macular retinal edema Diseases 0.000 title claims abstract description 33
- 206010030113 Oedema Diseases 0.000 claims abstract description 51
- 239000011159 matrix material Substances 0.000 claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 13
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 230000008859 change Effects 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 10
- 230000002207 retinal effect Effects 0.000 claims description 10
- 230000002146 bilateral effect Effects 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 238000005452 bending Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000000877 morphologic effect Effects 0.000 claims description 3
- 238000012014 optical coherence tomography Methods 0.000 abstract description 20
- 238000001514 detection method Methods 0.000 abstract description 3
- 230000003902 lesion Effects 0.000 abstract description 3
- 238000013399 early diagnosis Methods 0.000 abstract description 2
- 208000030533 eye disease Diseases 0.000 abstract description 2
- 239000010410 layer Substances 0.000 description 41
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 4
- 230000001575 pathological effect Effects 0.000 description 4
- 238000002310 reflectometry Methods 0.000 description 4
- 230000004256 retinal image Effects 0.000 description 4
- 238000009499 grossing Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 206010012689 Diabetic retinopathy Diseases 0.000 description 1
- 206010061218 Inflammation Diseases 0.000 description 1
- 208000017442 Retinal disease Diseases 0.000 description 1
- 206010064930 age-related macular degeneration Diseases 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000001125 extrusion Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000001727 in vivo Methods 0.000 description 1
- 230000008595 infiltration Effects 0.000 description 1
- 238000001764 infiltration Methods 0.000 description 1
- 230000004054 inflammatory process Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 208000002780 macular degeneration Diseases 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008397 ocular pathology Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 210000001957 retinal vein Anatomy 0.000 description 1
- 208000004644 retinal vein occlusion Diseases 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 210000004127 vitreous body Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10101—Optical tomography; Optical coherence tomography [OCT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Eye Examination Apparatus (AREA)
Abstract
The invention discloses an automatic retina layer segmentation method for an OCT (optical coherence tomography) image of macular edema, which comprises the following steps of: step 1, collecting a retina image with macular edema; step 2, preprocessing the image; step 3, constructing a weight matrix of the segmentation boundary; step 4, obtaining the shortest path of the retina boundary based on the shortest path fast algorithm; step 5, detecting and positioning an edema area; step 6, correcting error segmentation of the edema area; step 7, adopting a smooth boundary line; and 8, performing inverse flattening, namely performing automatic retina layer segmentation based on improved graph search, and automatically correcting error segmentation of an edema area based on edema detection and interpolation to quickly and accurately obtain 8 retina layer boundaries of the macular edema image. The invention provides an automatic noninvasive tool for analyzing the position of macular edema lesion and measuring and calculating the thickness of the retina layer, and simultaneously provides technical reference for researching the early diagnosis of macular edema related eye diseases.
Description
Technical Field
The invention relates to the field of image processing, in particular to an automatic retina layer segmentation method for macular edema Optical Coherence Tomography (OCT) images.
Background
OCT is an imaging technique for tomographic scanning of biological tissue using low coherence interference light. It has been widely used in ophthalmic diagnosis due to its advantages of high detection sensitivity, high resolution and non-invasiveness in vivo. Macular Edema (ME) is a common pathological result of many retinal diseases such as diabetes, age-related macular degeneration, retinal vein occlusion, and the like. Layer structure analysis of OCT retinal images in the presence of macular edema can reveal ocular pathology of the associated disease.
The retinal layer boundary segmentation methods of the prior art can be classified into the following 3 types: (1) the method based on graph search comprises the following steps: for example, in patent CN103514605, the method uses a shortest path algorithm Dijkstra to segment the boundary of the retinal layer based on the vertical gradient feature of the OCT retinal image; (2) method based on machine learning: and constructing a random forest classifier, learning boundary pixels between layers, obtaining final boundary probability segmentation through pixel classification, or extracting the characteristics of a specific retina layer boundary by utilizing a deep learning network to train a corresponding classifier to carry out layer estimation on the retina. (3) The method combines machine learning and graph theory, firstly obtains a probability graph of pixel classification through machine learning, and then obtains final accurate segmentation through graph theory.
In patent CN103514605, first, ILM layer positioning is performed according to the reflectivity difference between the vitreous body and the retina, then a similar high reflectivity RNFL layer is found according to the high reflectivity of ILM, then the RPE layer is estimated by using the structural characteristics of the high reflectivity retina, and finally, the retina layer segmentation is realized by Dijkstra algorithm by combining the positioned position and taking gradient information as a weight value.
With the increase of population and the rapid change of population structure to aging, the diseases of eyes are more and more common, wherein macular edema refers to the severe vision decline caused by the inflammatory reaction and liquid infiltration of the macular area of the most sensitive part of the fundus retina to light, which is the central retinal vein obstruction and diabetic retinopathy and is one of the important reasons causing the vision decline, and the edema area has structural influence on the retina layers.
Common graph search-based methods rely on gradient information at layer boundaries and may not achieve accurate segmentation due to extremely low inter-layer contrast or vessel shadowing. The machine learning-based method estimates the position of the layer boundary by extracting the features of a specific retina layer boundary and training a corresponding classifier, and a large amount of data is required in the process as a support for model training, which causes a large computational burden. In addition, these methods are basically designed for normal retinal images. Although the image with macular edema can be migrated to a certain degree, the edema can cause fusion between retina layers and cause extrusion on other layers, so that the segmentation effect of the edema area of the methods is not ideal, and manual correction is needed subsequently. In addition, in the current automatic retina layer segmentation method, the pathological condition retina layers are rarely segmented, and cannot be effectively segmented.
Disclosure of Invention
The invention overcomes the defects in the prior art and provides an automatic retina layer segmentation method for an OCT image of macular edema. The method provided by the invention provides an automatic noninvasive tool for analyzing the position of macular edema lesion and measuring and calculating the thickness of the retina layer, and simultaneously provides technical reference for researching the early diagnosis of macular edema related eye diseases.
In order to solve the technical problems, the invention is realized by the following technical scheme:
an automatic retina layer segmentation method for macular edema OCT images comprises the following steps:
step 1, collecting a retina image with macular edema;
step 2, preprocessing the image;
step 3, constructing a weight matrix of the segmentation boundary;
step 4, obtaining the shortest path of the retina boundary based on the shortest path fast algorithm;
step 5, detecting and positioning an edema area;
step 6, correcting error segmentation of the edema area;
step 7, adopting a smooth boundary line;
and 8, reversely flattening.
Further, the preprocessing the image in the step 2 comprises the following steps:
(1) speckle noise in the OCT image is suppressed by using self-adaptive weighted bilateral filtering, and main texture information of a retina layer is reserved;
(2) flattening the filtered retina image by using the BM boundary as a reference line, and reducing errors caused by obvious bending of the boundary;
(3) enhancing the contrast between the bright and dark layers, performing Gaussian filtering on the flattened b-scan for the first time, and setting the intensity value with the intensity smaller than the median value of the corresponding column as zero to perform threshold value on the image;
(4) b, normalizing the scanning to ensure that the pixel intensity is between 0 and 1;
further, the constructing of the weight matrix of the segmentation boundary in step 3 includes the following steps:
(1) solving a gradient map; because the change between the retina layers presents two forms of dark to light and light to dark, a dark to light gradient map A and a light to dark gradient map B are generated by adopting a formula (1), and the enhanced gradient maps are subjected to gamma conversion enhancement;
wherein I (x, y) is the pixel intensity value of the position (x, y), m is the number of pixels of the image in the x direction, and n is the number of pixels of the image in the y direction;
(2) defining a weight function; taking each pixel point in the image as a node, taking each image as a node graph, combining two constraint conditions of a vertical gradient change value and a coordinate change absolute value in the z direction, and defining a weight function between the nodes a and b as follows:
Wab＝0.8*(Ga+Gb)+0.2*Sab+Wmin (2)
wherein WabIs the weight of the edge connecting node a and node b, GaAnd GbVertical gradients at node a and node b, S, respectivelyabIs the absolute value of the coordinate change of the node a and the node b in the y direction, WminTo a minimumWeight is set to 10-5；
(3) The gradient map A and the weighting function are used for constructing a dark-channel light weighting matrix of ILM, INL-OPL, IS-OS, OS-RPE, and the gradient map B and the weighting function are used for constructing a light-to-dark weighting matrix of NFL-GCL, OPL-ONL, OS-RPE, BM.
Further, the step 4 is to obtain the shortest path of 8 boundaries of the retina based on a shortest path fast algorithm.
Further, the step 4 is to call the corresponding weight matrix and limit the region of interest, and sequentially divide by using an SPFA algorithm: ILM, IS-OS, BM, OS-RPE, NFL-GCL, OPL-ONL, IPL-INL, INL-OPL.
Further, the step 5 of detecting and locating the edema area comprises the following steps:
(1) defining a probability edema area by using pixels with pixel intensity values lower than the empirical value 20, displaying an image of the probability edema area by using a binary image, wherein a white pixel with a value of 1 represents the probability area, and a black pixel with a value of 0 represents the background; (2) setting an area of the edema area to be greater than 100 pixels; (3) applying morphological operations to the image to make the edema zone smoother; (4) the column coordinates of the edema are returned.
Further, the erroneous segmentation of the corrected edema area of step 6 is: according to the coordinates of the edema rows, the segmentation results of the edema rows are covered, and a one-dimensional linear interpolation method is adopted to enable the segmentation lines to directly pass through the edema, so that the automatic correction of the retinal layer boundary of the edema area is realized.
Further, the inverse flattening of step 8 is to inverse flatten the image by moving pixels up or down in a direction opposite to the image flattening direction, restoring the original curvature.
Compared with the prior art, the invention has the beneficial effects that:
the invention relates to an automatic retina layer segmentation method for an OCT (optical coherence tomography) image of macular edema, which can automatically segment and correct a retina image with edema lesion and effectively overcome the defects of time consumption and subjectivity of manual segmentation and manual correction.
The invention carries out retina layer segmentation based on improved graph search, wherein the weight is set to combine two constraint conditions of the vertical gradient value and the coordinate change absolute value of the image, thereby effectively reducing the influence of the vessel shadow on the segmentation and improving the accuracy of the segmentation.
The method does not need to train a model, and adopts a Shortest Path Fast Algorithm (SPFA) algorithm to replace the commonly used Dijkstra, thereby effectively improving the efficiency of segmenting the boundary of the retina layer.
The invention adds the edema detection and interpolation method to automatically correct the error segmentation of the edema area, effectively reduces the influence of edema on the retina layer segmentation and realizes the retina layer segmentation of normal and pathological states.
Drawings
The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, with the understanding that the present disclosure is to be considered as an exemplification of the invention and is not intended to limit the invention to the embodiments illustrated in the drawings, in which:
fig. 1 is a flowchart of an automatic retina layer segmentation method for macular edema OCT images according to the present invention.
Fig. 2 is a flow chart of image preprocessing.
FIG. 3 is a flow chart of constructing a weight matrix for segmenting boundaries.
Fig. 4 is a flow chart for detecting and locating an edema zone.
FIG. 5 is a diagram of non-foveated and foveated artwork and results of adaptive bilateral filtering.
Fig. 6 is a diagram showing the result of flattening the original image by BM boundaries.
Fig. 7 is a schematic diagram of an OCT retinal image of macular edema. The 8 layer boundaries are labeled and named, the upper left corner is the direction of the image in the coordinate system, and the dark low-reflectance region appearing in the middle of the retina is the edema.
Fig. 8 is a diagram showing the result of image preprocessing.
Fig. 9A is a dark to light gradient plot.
Fig. 9B is a light to dark gradient plot.
Fig. 10 is a diagram showing the results of the 8 layer boundaries obtained after automatic segmentation and rectification.
FIG. 11 is a schematic diagram of the final result after post-processing, wherein the post-processing comprises two steps of curve smoothing and counter-pressure smoothing.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Referring to fig. 1 to 11, the method for automatically segmenting the retinal layer of an OCT image of macular edema according to the present invention includes the following steps:
step 1, collecting a retina image with macular edema from a Topcon 3D-OCT instrument.
Step 2, preprocessing the image, referring to fig. 2, specifically:
(1) speckle noise in the OCT image is suppressed by using self-adaptive weighted bilateral filtering, and main texture information of a retina layer is reserved; speckle noise is easily caused due to the limited spatial frequency bandwidth in the OCT measurement, and the speckle noise greatly interferes and affects the contrast of an image, thereby causing an image segmentation error. However, because speckle noise belongs to strong noise, the traditional bilateral filtering method cannot remove the strong noise, because the difference between the pixel of the strong noise and the remaining pixel values in the region of the image is large, and the filtering is limited by the weight of the gray domain, the effect of the noise with large intensity after being filtered is not obvious. Adaptive weighted bilateral filtering can improve this drawback and effectively smooth the OCT image and maintain the texture information of the retinal layer, as shown in fig. 5.
(2) Flattening the filtered retina image by using the BM boundary as a reference line, and reducing errors caused by obvious bending of the boundary; the reason why the BM boundary is selected as the reference line is that BM regions have few pathological conditions and the gradient information is relatively obvious, which facilitates the domain search of the shortest path, as shown in fig. 6.
(3) Enhancing the contrast between the bright and dark layers, performing Gaussian filtering on the flattened b-scan for the first time, and setting the intensity value with the intensity smaller than the median value of the corresponding column as zero to perform threshold value on the image;
(4) the b-scan is normalized to ensure that the pixel intensity is between 0 and 1. In general, for gray scale images (or each color component of a color channel), the pixel gray scale values are distributed between 0 and 255, and normalization can further improve the image contrast, thereby facilitating subsequent processing.
Step 3, constructing a weight matrix of the segmentation boundary, referring to fig. 3, specifically:
(1) solving a gradient map; because the change between the retina layers takes two forms of dark to light and light to dark, the invention adopts the formula (1) to generate a dark to light gradient map A and a light to dark gradient map B, and carries out gamma conversion enhancement on the enhanced gradient maps.
Where I (x, y) is the pixel intensity value for location (x, y), m is the number of pixels in the x-direction of the image, and n is the number of pixels in the y-direction of the image.
(2) Defining a weight function; and taking each pixel point in the image as a node, and taking each image as a node graph. Combining two constraints of the vertical gradient change value and the coordinate change absolute value in the z direction, a weight function between the nodes a and b is defined as follows:
Wab＝0.8*(Ga+Gb)+0.2*Sab+Wmin (2)
wherein WabIs the weight of the edge connecting node a and node b, GaAnd GbVertical gradients at node a and node b, S, respectivelyabIs the absolute value of the coordinate change of the node a and the node b in the y direction, WminIs set to 10 for minimum weight-5。
(3) The gradient map A and the weighting function are used for constructing a dark-channel light weighting matrix of ILM, INL-OPL, IS-OS, OS-RPE, and the gradient map B and the weighting function are used for constructing a light-to-dark weighting matrix of NFL-GCL, OPL-ONL, OS-RPE, BM.
And 4, obtaining the shortest path of 8 boundaries of the retina based on a Shortest Path Fast Algorithm (SPFA), and referring to fig. 7. The method specifically comprises the following steps: calling a corresponding weight matrix and limiting the region of interest, and sequentially dividing by using an SPFA algorithm: ILM, IS-OS, BM, OS-RPE, NFL-GCL, OPL-ONL, IPL-INL, INL-OPL.
Step 5, detecting and locating the edema zone, see fig. 4. The method specifically comprises the following steps: (1) pixels with pixel intensity values below the empirical value of 20 define regions of probabilistic edema. Displaying an image of the probability edema area using a binary image, white pixels having a value of 1 representing the probability area, and black pixels having a value of 0 representing the background; (2) the area of the edema area is set to be greater than 100 pixels. We set 100 as the threshold because we found that the region of edema that affects the segmentation accuracy is typically greater than 100 pixels; (3) applying morphological operations to the image to make the edema zone smoother; (4) the column coordinates of the edema are returned.
And 6, correcting the wrong segmentation of the edema area. The method specifically comprises the following steps: according to the coordinates of the edema rows, the segmentation results of the edema rows are covered, and a one-dimensional linear interpolation method is adopted to enable the segmentation lines to directly pass through the edema, so that the automatic correction of the retinal layer boundary of the edema area is realized.
And 7, smoothing the boundary line of 8 edges by adopting a cubic spline interpolation method.
And 8, reversely flattening. The original curvature is restored by de-flattening the image by moving the pixels up or down in the direction opposite to the image flattening direction.
Finally, it should be noted that: although the present invention has been described in detail with reference to the embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments or portions thereof without departing from the spirit and scope of the invention.
Claims (9)
1. An automatic retina layer segmentation method for an OCT image of macular edema is characterized by comprising the following steps of:
step 1, collecting a retina image with macular edema;
step 2, preprocessing the image;
step 3, constructing a weight matrix of the segmentation boundary;
step 4, obtaining the shortest path of the retina boundary based on the shortest path fast algorithm;
step 5, detecting and positioning an edema area;
step 6, correcting error segmentation of the edema area;
step 7, adopting a smooth boundary line;
and 8, reversely flattening.
2. The method for automatically segmenting retina layers aiming at the macular edema OCT image as recited in claim 1, wherein the preprocessing the image of the step 2 comprises the following steps:
(1) speckle noise in the OCT image is suppressed by using self-adaptive weighted bilateral filtering, and main texture information of a retina layer is reserved;
(2) flattening the filtered retina image by using the BM boundary as a reference line, and reducing errors caused by obvious bending of the boundary;
(3) enhancing the contrast between the bright and dark layers, performing Gaussian filtering on the flattened b-scan for the first time, and setting the intensity value with the intensity smaller than the median value of the corresponding column as zero to perform threshold value on the image;
(4) the b-scan is normalized to ensure that the pixel intensity is between 0 and 1.
3. The method for automatically segmenting the retina layers of an OCT image of macular edema according to claim 1, wherein the step 3 of constructing the weight matrix of the segmentation boundary comprises the following steps:
(1) solving a gradient map; because the change between the retina layers presents two forms of dark to light and light to dark, a dark to light gradient map A and a light to dark gradient map B are generated by adopting a formula (1), and the enhanced gradient maps are subjected to gamma conversion enhancement;
wherein I (x, y) is the pixel intensity value of the position (x, y), m is the number of pixels of the image in the x direction, and n is the number of pixels of the image in the y direction;
(2) defining a weight function; taking each pixel point in the image as a node, taking each image as a node graph, combining two constraint conditions of a vertical gradient change value and a coordinate change absolute value in the z direction, and defining a weight function between the nodes a and b as follows:
Wab＝0.8*(Ga+Gb)+0.2*Sab+Wmin (2)
wherein WabIs the weight of the edge connecting node a and node b, GaAnd GbVertical gradients at node a and node b, S, respectivelyabIs the absolute value of the coordinate change of the node a and the node b in the y direction, WminIs set to 10 for minimum weight-5；
(3) The gradient map A and the weighting function are used for constructing a dark-channel light weighting matrix of ILM, INL-OPL, IS-OS, OS-RPE, and the gradient map B and the weighting function are used for constructing a light-to-dark weighting matrix of NFL-GCL, OPL-ONL, OS-RPE, BM.
4. The method for automatically segmenting the retina layers of the macular edema OCT image according to claim 1, wherein the step 4 is to obtain the shortest path of 8 boundaries of the retina based on the shortest path fast algorithm.
5. The method for automatically segmenting the retinal layer of the macular edema OCT image of claim 4, wherein the step 4 is to call the corresponding weight matrix and limit the interested area, and sequentially segment by using SPFA algorithm: ILM, IS-OS, BM, OS-RPE, NFL-GCL, OPL-ONL, IPL-INL, INL-OPL.
6. The method for automatically segmenting retina layers aiming at the macular edema OCT image as recited in claim 1, wherein the step 5 of detecting and locating the edema area comprises the following steps:
(1) defining a probability edema area by using pixels with pixel intensity values lower than the empirical value 20, displaying an image of the probability edema area by using a binary image, wherein a white pixel with a value of 1 represents the probability area, and a black pixel with a value of 0 represents the background; (2) setting an area of the edema area to be greater than 100 pixels; (3) applying morphological operations to the image to make the edema zone smoother; (4) the column coordinates of the edema are returned.
7. The method for automatically segmenting retina layers aiming at the macular edema OCT image as recited in claim 1, wherein the error segmentation of the corrected edema area of the step 6 is as follows: according to the coordinates of the edema rows, the segmentation results of the edema rows are covered, and a one-dimensional linear interpolation method is adopted to enable the segmentation lines to directly pass through the edema, so that the automatic correction of the retinal layer boundary of the edema area is realized.
8. The method for automatically segmenting the retinal layer of the macular edema OCT image of claim 1, wherein the step 7 uses cubic spline interpolation to smooth 8 boundary lines.
9. The method for automatically segmenting retinal layers of an OCT image of macular edema according to claim 1, wherein the de-flattening of step 8 is a de-flattening of the image by moving pixels up or down in a direction opposite to the image flattening direction, restoring the original curvature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111241258.7A CN114092405A (en) | 2021-10-25 | 2021-10-25 | Retina layer automatic segmentation method for macular edema OCT image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111241258.7A CN114092405A (en) | 2021-10-25 | 2021-10-25 | Retina layer automatic segmentation method for macular edema OCT image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114092405A true CN114092405A (en) | 2022-02-25 |
Family
ID=80297785
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111241258.7A Pending CN114092405A (en) | 2021-10-25 | 2021-10-25 | Retina layer automatic segmentation method for macular edema OCT image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114092405A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740099A (en) * | 2023-08-15 | 2023-09-12 | 南京博视医疗科技有限公司 | OCT image segmentation method and device |
| CN116957857A (en) * | 2023-09-19 | 2023-10-27 | 中国建筑西南设计研究院有限公司 | Building restoration method and device and electronic equipment |
-
2021
- 2021-10-25 CN CN202111241258.7A patent/CN114092405A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740099A (en) * | 2023-08-15 | 2023-09-12 | 南京博视医疗科技有限公司 | OCT image segmentation method and device |
| CN116740099B (en) * | 2023-08-15 | 2023-11-14 | 南京博视医疗科技有限公司 | OCT image segmentation method and device |
| CN116957857A (en) * | 2023-09-19 | 2023-10-27 | 中国建筑西南设计研究院有限公司 | Building restoration method and device and electronic equipment |
| CN116957857B (en) * | 2023-09-19 | 2024-01-16 | 中国建筑西南设计研究院有限公司 | Building restoration method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109493954B (en) | SD-OCT image retinopathy detection system based on category distinguishing and positioning | |
| Shi et al. | Automated 3-D retinal layer segmentation of macular optical coherence tomography images with serous pigment epithelial detachments | |
| WO2022007352A1 (en) | Three-dimensional choroidal vessel imaging and quantitative analysis method and apparatus based on optical coherence tomography system | |
| Soorya et al. | An automated and robust image processing algorithm for glaucoma diagnosis from fundus images using novel blood vessel tracking and bend point detection | |
| Hassan et al. | Deep learning based joint segmentation and characterization of multi-class retinal fluid lesions on OCT scans for clinical use in anti-VEGF therapy | |
| Ghorbel et al. | Automated segmentation of macular layers in OCT images and quantitative evaluation of performances | |
| US11854199B2 (en) | Methods and systems for ocular imaging, diagnosis and prognosis | |
| US20150371400A1 (en) | Segmentation and identification of closed-contour features in images using graph theory and quasi-polar transform | |
| CN114092405A (en) | Retina layer automatic segmentation method for macular edema OCT image | |
| Nyúl | Retinal image analysis for automated glaucoma risk evaluation | |
| Ouyang et al. | Accurate tissue interface segmentation via adversarial pre-segmentation of anterior segment OCT images | |
| US20210272291A1 (en) | Method and computer program for segmentation of optical coherence tomography images of the retina | |
| CN112330638B (en) | Retina OCT (optical coherence tomography) image horizontal registration and image enhancement method | |
| David et al. | A Comprehensive Review on Partition of the Blood Vessel and Optic Disc in RetinalImages | |
| WO2016108755A1 (en) | Method and apparatus for aligning a two-dimensional image with a predefined axis | |
| Akram et al. | Blood vessel enhancement and segmentation using wavelet transform | |
| Dikkala et al. | A comprehensive analysis of morphological process dependent retinal blood vessel segmentation | |
| WO2022007353A1 (en) | Method and apparatus for enhancing choroid oct image on basis of signal reverse compensation | |
| Sun et al. | 3D automatic segmentation method for retinal optical coherence tomography volume data using boundary surface enhancement | |
| Lejoyeux et al. | En-face analysis of short posterior ciliary arteries crossing the sclera to choroid using wide-field swept-source optical coherence tomography | |
| CN110033496B (en) | Motion artifact correction method for time sequence three-dimensional retina SD-OCT image | |
| Shi et al. | Automated choroid segmentation in three-dimensional 1-μ m wide-view OCT images with gradient and regional costs | |
| Pan et al. | Segmentation guided registration for 3d spectral-domain optical coherence tomography images | |
| CN109919098B (en) | Target object identification method and device | |
| Shabbir et al. | A comparison and evaluation of computerized methods for blood vessel enhancement and segmentation in retinal images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |