CN113393478A - OCT retina layering method, system and medium based on convolutional neural network - Google Patents
OCT retina layering method, system and medium based on convolutional neural network Download PDFInfo
- Publication number
- CN113393478A CN113393478A CN202110556870.7A CN202110556870A CN113393478A CN 113393478 A CN113393478 A CN 113393478A CN 202110556870 A CN202110556870 A CN 202110556870A CN 113393478 A CN113393478 A CN 113393478A
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- path
- image
- oct
- 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
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 83
- 238000000034 method Methods 0.000 title claims abstract description 67
- 210000001525 retina Anatomy 0.000 title claims abstract description 53
- 230000002207 retinal effect Effects 0.000 claims abstract description 46
- 238000013528 artificial neural network Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 15
- 230000004913 activation Effects 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 abstract description 25
- 230000008569 process Effects 0.000 abstract description 3
- 238000012014 optical coherence tomography Methods 0.000 description 66
- 239000010410 layer Substances 0.000 description 29
- 230000011218 segmentation Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 6
- 210000004126 nerve fiber Anatomy 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 230000036541 health Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 239000012528 membrane Substances 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000004256 retinal image Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000010349 pulsation Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 238000013399 early diagnosis Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000003707 image sharpening Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000010827 pathological analysis Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
- 238000012795 verification Methods 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
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Abstract
The invention discloses an OCT retina layering method, a system and a medium based on a convolutional neural network, wherein the OCT retina layering method based on the convolutional neural network comprises the following steps: acquiring an image dataset consisting of retinal OCT images; segmenting a standard image dataset from the image dataset using a shortest path algorithm; inputting the standard image data set into a convolutional neural network as a standard for training to obtain a trained first neural network; and inputting all data sets except the standard image data set in the image data set into the first neural network to obtain a retina layering result output by the first neural network. The invention can greatly improve the data processing speed, improve the data processing efficiency, facilitate the processing of a large amount of data, avoid a large amount of personnel training processes and reduce the cost.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an OCT retina layering method, system and medium based on a convolutional neural network.
Background
The thickness of the retinal nerve fiber layer is closely related to eye health, the thickness of the retinal nerve fiber layer reflects the eye health progress change, and the pulsation parameter of the retinal nerve fiber layer can provide a basis for early diagnosis of eyes, so that the method has important significance for layered extraction research of the retina.
The retinal OCT image layering method in the related technology is mainly divided into the following two categories:
the first type: the medical professionals manually segment the retina images, and after the OCT equipment collects the fundus retina images, the professional medical knowledge personnel manually segment the retina nerve fiber layers according to experience for pathological analysis.
The second type: the traditional image segmentation method mainly carries out edge detection and feature extraction on an image through a traditional feature extraction algorithm. The generalization of the algorithm is not ideal under the condition of a large amount of data, more prior knowledge is needed for medical images, and the cost is higher.
Disclosure of Invention
The present invention is directed to solving at least one of the problems of the prior art. Therefore, the invention provides an OCT retina layering method based on a convolutional neural network, which can be used for layering a large number of retina pictures with different sizes and different positions by using a trained neural network, greatly improves the speed and accuracy of data processing, and has a good processing effect on a large amount of data.
The invention also provides an OCT retina layering system based on the convolutional neural network, which has the OCT retina layering method based on the convolutional neural network.
The invention also provides a computer readable storage medium.
In a first aspect, the present embodiment provides an OCT retinal layering method based on a convolutional neural network, including the following steps:
acquiring an image dataset consisting of an OCT (Optical Coherence Tomography) image of a retina;
segmenting a standard image dataset from the image dataset using a shortest path algorithm;
inputting the standard image data set into a convolutional neural network as a standard for training to obtain a trained first neural network;
and inputting all data sets except the standard image data set in the image data set into the first neural network to obtain a retina layering result output by the first neural network.
The OCT retina layering method based on the convolutional neural network has at least the following beneficial effects:
firstly, collecting a retina OCT image, processing the retina OCT image to form an image data set, segmenting a standard image data set from the image data set by using a shortest path algorithm, wherein the shortest path algorithm is a method for calculating the shortest path from one node to all other nodes, and can be used for layering the OCT retina image because the characteristics of all layers of the OCT retina image are obvious; taking a B-scan image of each frame of the SD-OCT as a node map, wherein each node corresponds to one pixel; links connecting nodes are called edges, a group of connected edges forms a path through the graph, and a priority shortest path of the path is created by assigning weights to the individual edges. The priority shortest path is a path with the minimum sum of total weights from a starting node to an ending node in all paths passing through the graph, and different areas of the image are divided through the shortest path; inputting the standard image data set into a convolutional neural network as a standard to be trained to obtain a first neural network, wherein the first neural network is the trained convolutional neural network; and inputting the acquired image data sets except the standard image data set into the first neural network for processing to obtain a retina layering result.
Compared with a conventional manual segmentation mode by medical professionals and a conventional image segmentation method, the OCT retina layering method based on the convolutional neural network provided by the embodiment greatly improves the data processing speed, improves the data processing efficiency, is more convenient to process a large amount of data, does not need a large amount of personnel training processes, and reduces the cost.
According to some embodiments of the invention, the segmenting the standard image dataset from the image dataset using a shortest path algorithm comprises the steps of:
acquiring a B-scan image of the retina;
calculating the gradient weight of the scanning image B in the vertical direction, and initializing the line addition at two sides of the layer end image;
limiting a search area, and setting the segmented nodes as invalid nodes;
summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a first path, and judging whether the first path is the shortest path or not;
and if the first path is the shortest path, outputting the pixel data set represented by the first path.
According to some embodiments of the invention, the convolutional neural network comprises an input layer, a convolutional layer, an activation function, a pooling layer, and a fully-connected layer.
According to some embodiments of the invention, the activation function is a step function, mapping an input value to 1 or 0, mapping an input value to 1 representing neuron excitation and mapping an input value to 0 representing neuron inhibition.
According to some embodiments of the invention, the calculating the vertical gradient weight of the B-scan image comprises:
taking each point pixel of the B scanning image as a node in the shortest path algorithm, and calculating the pixel intensity change gradient in the vertical direction, wherein the weight w between the node a and the node BabComprises the following steps:
wab＝2-(ga+gb)+wmin
gaand gbVertical gradient, w, of the image at nodes a and b, respectivelyminIs the smallest weight in the image.
According to some embodiments of the present invention, after the determining whether the first path is the shortest path, the method further includes:
if the first path is not the shortest path, summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a second path, and judging whether the second path is the shortest path or not;
and if the second path is the shortest path, outputting the pixel represented by the second path.
In a second aspect, the present embodiment provides an OCT retinal layering system based on a convolutional neural network, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the computer program implementing the convolutional neural network-based OCT retinal layering method according to the first aspect.
The OCT retina layering system based on the convolutional neural network has at least the following beneficial effects:
the OCT retina layering system based on the convolutional neural network applies the OCT retina layering method based on the convolutional neural network in the first aspect, so that the speed and the efficiency of data processing can be effectively improved, a good effect is achieved on processing of a large amount of data, meanwhile, the cost of personnel training is reduced, and the OCT retina layering system based on the convolutional neural network has good practical value.
In a third aspect, the present embodiments provide a computer-readable storage medium storing computer-executable instructions for causing a computer to perform the convolutional neural network-based OCT retinal layering method according to the first aspect.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which the abstract is to be fully consistent with one of the figures of the specification:
FIG. 1 is a flow chart of an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 2 is a flow chart of an OCT retinal layering method based on a convolutional neural network provided by another embodiment of the present invention;
FIG. 3 is a data set of an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 4 is a schematic diagram of a Dijkstra algorithm for determining a shortest path according to an OCT retina layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 5 is a table of weight values for an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 6 is a flowchart of an algorithm of an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 7 is an exemplary segmentation graph of initialization endpoints for an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the present invention;
FIG. 8 is a diagram of the retinal segmentation results of the OCT retinal layering method based on the convolutional neural network provided by one embodiment of the present invention;
FIG. 9 is a diagram of a convolutional neural network structure of an OCT retinal layering method based on a convolutional neural network provided by an embodiment of the invention;
FIG. 10 is a graph of the output of a convolutional neural network of the OCT retinal layering method based on a convolutional neural network provided by one embodiment of the present invention;
FIG. 11 is a graph of the output of a convolutional neural network of the OCT retinal layering method based on a convolutional neural network provided by one embodiment of the present invention;
fig. 12 is a graph of the output result of the convolutional neural network of the OCT retinal layering method based on the convolutional neural network according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
It should be noted that although functional block divisions are provided in the system drawings and logical orders are shown in the flowcharts, in some cases, the steps shown and described may be performed in different orders than the block divisions in the systems or in the flowcharts. The terms first, second and the like in the description and in the claims, and the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.
The thickness of the retinal nerve fiber layer reflects the eye health, and the pulsation parameters of the retinal nerve fiber layer can provide reference for the eye health, so that the method has great significance for the layered extraction research of the retina.
The invention provides an OCT retina layering method based on a convolutional neural network, which can be used for layering a large number of retina pictures with different sizes and different positions by using a trained neural network, greatly improves the speed and the accuracy of data processing, and has a good processing effect on a large number of data.
The embodiments of the present invention will be further explained with reference to the drawings.
Referring to fig. 1 and 2, fig. 1 is a flowchart of an OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention, and fig. 2 is a flowchart of an OCT retinal layering method based on a convolutional neural network according to another embodiment of the present invention, where the OCT retinal layering method based on a convolutional neural network includes, but is not limited to, steps S110 to S140.
Step S110, acquiring an image data set consisting of retina OCT images;
step S120, segmenting a standard image data set from the image data set by using a shortest path algorithm;
step S130, inputting the standard image data set into a convolutional neural network as a standard for training to obtain a first neural network after training is finished;
step S140, inputting all data sets except the standard image data set in the image data set into the first neural network, and obtaining a retinal layering result output by the first neural network.
Firstly, collecting a retina OCT image, processing the retina OCT image to form an image data set, and segmenting a group of image data sets by using a shortest path algorithm to obtain a standard image data set, wherein the shortest path algorithm is a method for calculating the shortest path from one node to all other nodes; taking a B-scan image of each frame of the SD-OCT as a node map, wherein each node corresponds to one pixel; links connecting nodes are called edges, a group of connected edges forms a path through the graph, and a priority shortest path of the path is created by assigning weights to the individual edges. The priority shortest path is a path with the minimum sum of total weights from a starting node to an ending node in all paths passing through the graph, and different areas of the image are divided through the shortest path; inputting the standard image data set serving as a standard into a convolutional neural network for training to obtain a first neural network, wherein the first neural network is a trained convolutional neural network; and inputting the acquired image data sets except the standard image data set into a first neural network for processing to obtain a retina layering result.
Referring to fig. 3, 4 and 5, fig. 3 is a data set of an OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention, fig. 4 is a schematic diagram of a Dijkstra algorithm for determining a shortest path in the OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention, and fig. 5 is a weight value table of the OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention.
In one embodiment, a retina OCT image to be detected is acquired from an OCT image equipment instrument and an image data set is established, a group of image data sets are segmented by using a shortest path algorithm to obtain a standard image data set, and then the standard image data set is input into a convolutional neural network as a standard to be trained to obtain a first neural network, wherein the first neural network is a trained convolutional neural network; and inputting the acquired image data sets except the standard image data set into the first neural network for processing to obtain a retina layering result.
The shortest path algorithm is a method of calculating the shortest path from one node to all other nodes. Since the various interlayer features of the OCT retinal image are significant, the shortest path algorithm can be used for the layering of the OCT retinal image. The B-scan image of SD-OCT per frame is taken as a node map, where each node corresponds to one pixel. Links connecting nodes are called edges, a group of connected edges forms a path through the graph, and a priority shortest path of the path is created by assigning weights to the individual edges. The priority shortest path is a path having the smallest sum of total weights from a start node to an end node among all paths passing through the graph, and different regions of the image are divided by the shortest path.
The shortest path is determined by using Dijkstra (Dijkstra) algorithm, three connected nodes are shown in fig. 4, fig. 5 is a weight value table among the three nodes, wherein one axis represents a starting node, the other axis represents an ending node, a minimum weighted path is to be found by using the Dijkstra algorithm, the weight value is required to be positive and ranges from 0 to 1, wherein the weight value of an edge is 0 and represents an unconnected node pair, and it can be seen from the weight value table that the weight value of 0.2 of node 1 to node 3 is lower than the weight value of 0.5 of node 1 to node 2, so that the path from node 1 to node 3 is the shortest path.
Referring to fig. 6, fig. 6 is a flowchart of an algorithm of an OCT retina layering method based on a convolutional neural network according to an embodiment of the present invention.
In an embodiment, segmenting a set of image datasets using a shortest path algorithm results in a standard image dataset comprising the steps of:
acquiring a B-scan image of the retina;
calculating the gradient weight of the scanning image B in the vertical direction, and initializing the line addition at the two sides of the end point image of the image layer;
limiting a search area, and setting the segmented nodes as invalid nodes;
summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a first path, and judging whether the first path is the shortest path or not;
if the first path is the shortest path, outputting the pixel represented by the first path.
Firstly, obtaining a B scanning image of a retina, wherein the B scanning image is a two-dimensional imaging scanning image, then calculating the gradient weight of the B scanning image in the vertical direction, and initializing the row addition at two sides of an end point image of a layer; calculating the weight, taking each point pixel in the input original image as a node in the shortest path algorithm, wherein the retinal image collected in the embodiment is mainly of a horizontal structure, the weight value in the shortest path algorithm is calculated by selecting the pixel intensity change gradient in the vertical direction, and the weight between the nodes a and b is as follows:
wab＝2-(ga+gb)+wmin
gaand gbVertical gradient, w, of the image at nodes a and b, respectivelyminAnd normalizing the gradients at each node to enable the values of the gradients to be between 0 and 1 for the minimum weight in the image and increasing a small positive value for stabilizing the system, wherein the normalized gradients serve as the weight of the shortest path algorithm to calculate the paths among the pixel points.
Initializing the end points of the image layers, respectively adding a row of nodes at two sides of the image because the retina to be segmented extends to occupy the whole width of the OCT image, and setting the weight values of the added row of nodes as the minimum weight values w in the imagemin，wminSignificantly less than any non-zero weight in the original graph, since Dijkstra's algorithm would select the weight and the path of least weight, while adding nodes in columns would allow the segmentation to traverse with little resistance in the vertical direction of the added column.
And limiting the search area, setting the divided nodes as invalid nodes and removing the weight of edges in the invalid nodes in order to avoid repeated search of the divided areas.
Searching a minimum weighted path, summing the weights around the starting node to obtain the node with the minimum weight, judging whether the found path is the shortest path or not, if not, searching in the adjacent area again, if so, outputting the pixel represented by the shortest path, thus the shortest path obtained according to the weights among all the pixel points is the required partition line.
Referring to fig. 7 and 8, fig. 7 is an initialization end point example segmentation graph of the OCT retinal layering method based on the convolutional neural network according to an embodiment of the present invention, and fig. 8 is a retinal segmentation result graph of the OCT retinal layering method based on the convolutional neural network according to an embodiment of the present invention.
Each black square represents a node, the color of the square represents weight, for the sake of clarity, the darker the color represents the smaller the weight, the square part is the original graph, the left and right columns are added columns, the weight of the node in the column is wminThe line from left to right through fig. 7 represents the shortest path found.
Referring to fig. 9, fig. 9 is a diagram of a convolutional neural network structure of an OCT retina layering method based on a convolutional neural network according to an embodiment of the present invention.
In one embodiment, a convolutional neural network includes an input layer, a convolutional layer, an activation function, a pooling layer, and a fully-connected layer. The convolutional neural network has a multilayer network structure and mainly comprises an input layer, a convolutional layer, an activation function, a pooling layer, a full-link layer and the like.
An input layer: mainly containing a large number of sets of picture data to be processed.
And (3) rolling layers: the convolution layer can extract various characteristics of the image by performing convolution on the input image through different convolution cores, and is applied to tasks such as image sharpening, edge extraction and segmentation.
Activation function: an ideal activation function is a step function, i.e. mapping the input value to 1 or 0, representing the neuron being excited or inhibited, respectively.
The method can be used for dividing 2000 retina OCT images with calibration data in a data set into three parts, namely a training part, a verification part and a testing part, wherein the ratio of the three parts is 8: 1: 1 is input into a convolutional neural network. The convolutional neural network comprises 15 convolutional layers of an input layer and Conv 1-5, the size of a convolutional kernel is 3 x 3, and a ReLU excitation function. Conv 1-5 convolution feature layer channel numbers are 32, 64, 128, 256 and 3 respectively.
Referring to fig. 10, 11 and 12, fig. 10 is a graph showing an output result of a convolutional neural network of an OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention, fig. 11 is a graph showing an output result of a convolutional neural network of an OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention, and fig. 12 is a graph showing an output result of a convolutional neural network of an OCT retinal layering method based on a convolutional neural network according to an embodiment of the present invention.
Fig. 10 is a segmented image in which the outputs of three convolutional neural networks are selected from the results, which may be the 40 th frame, the 84 th frame, and the 137 th frame, and fig. 10 also shows the inner limiting membrane and the outer limiting membrane.
The rest data are input into the neural network again, the output result is shown in fig. 11 and fig. 12, fig. 11 shows that the situation that the segmentation line cannot be accurately identified due to the incomplete edge of the right inner limiting membrane exists, in the subsequent processing, the part where the edge of the right inner limiting membrane is connected with the edge of the image is cut off, fig. 12 shows that the shadow of some small blood vessels in the image can be identified as the edge, and in the subsequent processing, a threshold value is set to remove the part jumping in the edge segmentation line.
In an embodiment, the activation function is a step function, that is, the input value is mapped to 1 or 0, which respectively represents that the neuron is excited or inhibited, and the save function can better reflect the state of the retina, thereby facilitating layering processing of the retina and having better practical effect.
In one embodiment, the weight value in the shortest path algorithm is selected and calculated by the gradient of pixel intensity variation in the vertical direction, and the weight between the nodes a and b is:
wab＝2-(ga+gb)+wmin
gaand gbVertical gradient, w, of the image at nodes a and b, respectivelyminFor the minimum weight in the image, a small positive number of values added to stabilize the system, the gradients at each node are normalized to have values between 0 and 1And taking the normalized gradient as the weight of the shortest path algorithm to calculate the path between the pixel points.
In an embodiment, after determining whether the first path is the shortest path, the method includes the steps of:
if the first path is not the shortest path, summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a second path, and judging whether the second path is the shortest path;
if the second path is the shortest path, the pixel represented by the second path is output.
The OCT retina layering method based on the convolutional neural network can effectively improve the speed and efficiency of data processing, has a good effect on processing a large amount of data, reduces the personnel training cost, and has good practical value.
The invention also provides an OCT retina layering system based on the convolutional neural network, which comprises: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the computer program implementing the above-described convolutional neural network-based OCT retinal layering method. The OCT retina layering system based on the convolutional neural network applies the OCT retina layering method based on the convolutional neural network, can effectively improve the speed and efficiency of data processing, has a good effect on processing of a large amount of data, reduces the cost of personnel training, and has good practical value.
Furthermore, an embodiment of the present invention also provides a computer-readable storage medium storing computer-executable instructions for execution by one or more control processors, e.g., the control processors are capable of performing method steps S110-S140 in fig. 1.
One of ordinary skill in the art will appreciate that all or some of the steps, systems, and methods disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an illustrative embodiment," "an example," "a specific example," or "some examples" or the like mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
While the preferred embodiments of the present invention have been described in detail, it will be understood by those skilled in the art that the foregoing and various other changes, omissions and deviations in the form and detail thereof may be made without departing from the scope of this invention.
Claims (8)
1. An OCT retina layering method based on a convolutional neural network is characterized by comprising the following steps:
acquiring an image dataset consisting of retinal OCT images;
segmenting a standard image dataset from the image dataset using a shortest path algorithm;
inputting the standard image data set into a convolutional neural network as a standard for training to obtain a trained first neural network;
and inputting all data sets except the standard image data set in the image data set into the first neural network to obtain a retina layering result output by the first neural network.
2. The convolutional neural network based OCT retinal layering method of claim 1, wherein the using a shortest path algorithm to segment a standard image dataset from the image dataset comprises the steps of:
acquiring a B-scan image of the retina;
calculating the gradient weight of the scanning image B in the vertical direction, and initializing the line addition at two sides of the layer end image;
limiting a search area, and setting the segmented nodes as invalid nodes;
summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a first path, and judging whether the first path is the shortest path or not;
and if the first path is the shortest path, outputting the pixel data set represented by the first path.
3. The convolutional neural network-based OCT retinal layering method of claim 1, wherein the convolutional neural network comprises an input layer, a convolutional layer, an activation function, a pooling layer, and a fully-connected layer.
4. The convolutional neural network-based OCT retinal layering method of claim 3, wherein the activation function is a step function.
5. The convolutional neural network-based OCT retinal layering method of claim 2, wherein the calculating the vertical direction gradient weight of the B-scan image comprises the steps of:
taking each point pixel of the B scanning image as a node in the shortest path algorithm, and calculating the pixel intensity change gradient in the vertical direction, wherein the weight w between the node a and the node BabComprises the following steps:
wab＝2-(ga+gb)+wmin
gaand gbVertical gradient, w, of the image at nodes a and b, respectivelyminIs the smallest weight in the image.
6. The OCT retinal layering method based on convolutional neural network of claim 2, further comprising the step of, after said determining whether the first path is the shortest path:
if the first path is not the shortest path, summing the weights around the starting node in the search area, taking the node with the minimum weight to obtain a second path, and judging whether the second path is the shortest path or not;
and if the second path is the shortest path, outputting the pixel represented by the second path.
7. An OCT retinal layering system based on a convolutional neural network, comprising: memory, processor and computer program stored on the memory and executable on the processor, characterized in that the processor when executing the computer program implements the convolutional neural network-based OCT retinal layering method of any one of claims 1 to 6.
8. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the convolutional neural network-based OCT retinal layering method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110556870.7A CN113393478A (en) | 2021-05-21 | 2021-05-21 | OCT retina layering method, system and medium based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110556870.7A CN113393478A (en) | 2021-05-21 | 2021-05-21 | OCT retina layering method, system and medium based on convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113393478A true CN113393478A (en) | 2021-09-14 |
Family
ID=77618945
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110556870.7A Pending CN113393478A (en) | 2021-05-21 | 2021-05-21 | OCT retina layering method, system and medium based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113393478A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106408562A (en) * | 2016-09-22 | 2017-02-15 | 华南理工大学 | Fundus image retinal vessel segmentation method and system based on deep learning |
| US20180012359A1 (en) * | 2016-07-06 | 2018-01-11 | Marinko Venci Sarunic | Systems and Methods for Automated Image Classification and Segmentation |
| CN108765388A (en) * | 2018-05-17 | 2018-11-06 | 苏州大学 | The automatic division method and system of esophagus endoscopic OCT image level structure |
| CN109272507A (en) * | 2018-07-11 | 2019-01-25 | 武汉科技大学 | The layer dividing method of coherent light faultage image based on structure Random Forest model |
-
2021
- 2021-05-21 CN CN202110556870.7A patent/CN113393478A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180012359A1 (en) * | 2016-07-06 | 2018-01-11 | Marinko Venci Sarunic | Systems and Methods for Automated Image Classification and Segmentation |
| CN106408562A (en) * | 2016-09-22 | 2017-02-15 | 华南理工大学 | Fundus image retinal vessel segmentation method and system based on deep learning |
| CN108765388A (en) * | 2018-05-17 | 2018-11-06 | 苏州大学 | The automatic division method and system of esophagus endoscopic OCT image level structure |
| CN109272507A (en) * | 2018-07-11 | 2019-01-25 | 武汉科技大学 | The layer dividing method of coherent light faultage image based on structure Random Forest model |
Non-Patent Citations (3)
| Title |
|---|
| STEPHANIE J. CHIU ET AL.: "Automatic segmentation of seven retinal layers in SDOCT images congruent with expert manual segmentation", 《OPTICS EXPRESS》 * |
| 唐艳红 等: "基于CNN和改进的图搜索分割OCT图像中的视网膜层", 《激光与光电子学进展》 * |
| 沈彬伟: "基于卷积神经网络和图搜索的OCT图像视网膜层分割研究", 《中国优秀博硕士学位论文全文数据库(硕士) 医药卫生科技辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110197493B (en) | Fundus image blood vessel segmentation method | |
| CN110599451A (en) | Medical image focus detection positioning method, device, equipment and storage medium | |
| CN107644225A (en) | Pulmonary lesionses recognition methods, device and realization device | |
| CN109816666B (en) | Symmetrical full convolution neural network model construction method, fundus image blood vessel segmentation device, computer equipment and storage medium | |
| CN103026379A (en) | Method for estimating image noise level | |
| JP2019192215A (en) | 3d quantitative analysis of retinal layers with deep learning | |
| WO2022087853A1 (en) | Image segmentation method and apparatus, and computer-readable storage medium | |
| CN107204001B (en) | Automatic midium segmentation method in carotid artery ultrasonic image | |
| CN111612756B (en) | Coronary artery specificity calcification detection method and device | |
| WO2018049598A1 (en) | Ocular fundus image enhancement method and system | |
| JPWO2020022027A1 (en) | Learning device and learning method | |
| JP2018531773A (en) | System and method for segmentation of optical coherence tomography slices | |
| CN111179295B (en) | Improved two-dimensional Otsu threshold image segmentation method and system | |
| CA3104562A1 (en) | Method and computer program for segmentation of optical coherence tomography images of the retina | |
| WO2015143435A1 (en) | Graph search using non-euclidean deformed graph | |
| CN117058676B (en) | Blood vessel segmentation method, device and system based on fundus examination image | |
| CN113920109A (en) | Medical image recognition model training method, recognition method, device and equipment | |
| CN113658165A (en) | Cup-to-tray ratio determining method, device, equipment and storage medium | |
| JP6603814B2 (en) | Retina image processing | |
| CN113298800A (en) | Processing method, device and equipment of CT angiography CTA source image | |
| US20220301177A1 (en) | Updating boundary segmentations | |
| Bhuvaneswari et al. | Contrast enhancement of retinal images using green plan masking and whale optimization algorithm | |
| CN113393478A (en) | OCT retina layering method, system and medium based on convolutional neural network | |
| CN109919098A (en) | The recognition methods of target object and device | |
| CN111815569B (en) | Image segmentation method, device, equipment and storage medium based on deep learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210914 |
|
| RJ01 | Rejection of invention patent application after publication |