CN114359219A - OCT image layering and focus semantic segmentation method, device and storage medium - Google Patents

OCT image layering and focus semantic segmentation method, device and storage medium Download PDF

Info

Publication number
CN114359219A
CN114359219A CN202111675172.5A CN202111675172A CN114359219A CN 114359219 A CN114359219 A CN 114359219A CN 202111675172 A CN202111675172 A CN 202111675172A CN 114359219 A CN114359219 A CN 114359219A
Authority
CN
China
Prior art keywords
stage
semantic segmentation
layering
area
data
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
Application number
CN202111675172.5A
Other languages
Chinese (zh)
Inventor
陈新建
汪竟成
范煜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou Bigvision Medical Technology Co ltd
Original Assignee
Suzhou Bigvision Medical Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Suzhou Bigvision Medical Technology Co ltd filed Critical Suzhou Bigvision Medical Technology Co ltd
Priority to CN202111675172.5A priority Critical patent/CN114359219A/en
Publication of CN114359219A publication Critical patent/CN114359219A/en
Priority to PCT/CN2022/088759 priority patent/WO2023123765A1/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • 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
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Software Systems (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Geometry (AREA)
  • Eye Examination Apparatus (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses an OCT image layering and focus semantic segmentation method, a device and a storage medium, wherein the method comprises the following steps: acquiring image data of each BScan two-dimensional slice of the OCT of one eye of a patient to be inspected; inputting the slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layering, the optic disc region and the macular depression region in the first stage; according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the macular depression area in the first stage, a 'concerned area' is manufactured; acquiring a data map of a 'region of interest'; the data map of the 'concerned area' is input into the convolution neural network of the second stage to obtain the semantic segmentation prediction results of the second stage of layering, drusen, pigment epithelium layer separation and choroid neovascular lesions.

Description

OCT image layering and focus semantic segmentation method, device and storage medium
Technical Field
The invention relates to an OCT image layering and focus semantic segmentation method, device and storage medium, belonging to the technical field of semantic segmentation.
Background
Optical Coherence Tomography (OCT) is an imaging technique for forming images of the fundus oculi, and because it can reflect the reflection and scattering characteristics of different physiological structures of the fundus oculi on incident weak coherent light, the formed three-dimensional images have depth level information, which has unique advantages compared with images such as fundus color photography.
At present, the hierarchical analysis and lesion analysis of OCT images usually employ a deep convolutional neural network to perform semantic segmentation of the hierarchy and lesion on OCT slices or whole OCT, and then perform post-processing according to the segmentation result to obtain some physiological indexes of the fundus, such as the thickness of the pigment epithelium layer and the volume of drusen. Generally, a deep learning model needs to perform hierarchical/focal segmentation on pixels of all regions of an acquired OCT image in a training stage, and the learned image information lacks pertinence, so that it is difficult to accurately identify a focus such as a small-area focus (e.g., drusen) related to age-related macular degeneration, and difficulty is brought to subsequent stage classification of the age-related macular degeneration according to information such as the volume of the focus.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides an OCT image layering and focus semantic segmentation method, device and storage medium, which improve the proportion of focus pixels in total pixels and reduce the learning difficulty of small-area positive focuses such as drusen.
In order to achieve the purpose, the invention is realized by adopting the following technical scheme:
in a first aspect, the present invention provides an OCT image layering and lesion semantic segmentation method, including:
acquiring image data of each BScan two-dimensional slice of OCT of one eye of a patient to be inspected, and performing normalization processing;
inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layering, the optic disc region and the macular depression region in the first stage;
according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the macular depression area in the first stage, a 'concerned area' is manufactured;
acquiring a data map of the attention area, and carrying out normalization processing on the data;
and inputting the normalized data image of the attention area into the convolution neural network of the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus of the second stage.
Further, the step of creating a "region of interest" based on the obtained semantic segmentation prediction results of the first-stage layer, the optic disc region, and the macular pucker region includes:
selecting a slice with the thinnest thickness of a nerve fiber layer where the central point of a yellow spot depressed area is located in a data label of the same-eye OCT, wherein the central point of the yellow spot depressed area is used as the central point of the yellow spot fovea of the OCT image of the eye;
selecting the maximum value of the width of the optic disc area in all the section data labels of the same-eye OCT as the 'optic disc diameter' of the OCT image;
according to the horizontal and longitudinal physical resolution and pixel resolution values in the specific mode of the OCT model, converting the horizontal pixel number of 2 optic disc diameters into the horizontal physical size, determining a 'attention area circle' by taking the central concave point of the yellow spot as the center of a circle and the physical value of the optic disc diameter as the radius, and converting the horizontal pixel number into the pixel width of an 'attention area' corresponding to the 'attention area circle' in each slice;
and taking the column of the central foveal point of the macula as a central point on the section, obtaining the pixel width through conversion of the width, taking the center of a polygonal external rectangle of the fundus layer structure label non-background areas of the columns as the center, and taking the specific pixel number as the height, and obtaining the 'region of interest'.
Further, the training method of the convolutional neural network of the first stage includes:
acquiring BScan two-dimensional slice image data of OCT and corresponding semantic segmentation labels of layering, optic disc area and yellow spot sunken area, performing data augmentation and normalization, and randomly disordering to manufacture a data set of a first stage;
at each step of the first-stage network training, inputting image data processed in a data set at the first stage into a convolutional neural network at the first stage to obtain a semantic segmentation prediction result of the layering, the optic disc region and the macular depression region at the first stage, calculating loss with a corresponding label, and performing gradient back propagation and network parameter updating;
according to BScan two-dimensional slice image data of OCT and corresponding semantic segmentation labels of layering, optic disc area and yellow spot sunken area, making a 'key area';
acquiring a data map and a layered focus label map of a 'key region', drusen, pigment epithelial layer detachment and choroidal neovascularization, performing data amplification, normalizing and randomly disordering the data, and making into a data set of a second stage;
and in each step of the second stage of network training, inputting the image data processed in the second stage of data set into the convolutional neural network in the first stage to obtain the semantic segmentation prediction results of the layering, drusen, pigment epithelium layer separation and choroidal neovascular lesion in the second stage, calculating loss with the corresponding label, and performing gradient back propagation and network parameter updating.
Further, the method for making the key area is the same as the method for making the attention area.
Further, the method also comprises the following steps: and optimizing the network by adopting an Adam optimizer.
In a second aspect, the present invention provides an OCT image layering and lesion semantic segmentation apparatus, including:
the system comprises a first acquisition unit, a second acquisition unit and a third acquisition unit, wherein the first acquisition unit is used for acquiring each BScan two-dimensional slice image data of the OCT of one eye of a patient to be inspected and carrying out normalization processing;
the first input unit is used for inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layers, the optic disc region and the macular depression region in the first stage;
the processing unit is used for manufacturing a focus area according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the yellow spot sunken area in the first stage;
the second acquisition unit is used for acquiring a data map of the attention area and normalizing the data;
and the second input unit is used for inputting the normalized data image of the attention area into the convolution neural network at the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus at the second stage.
In a third aspect, the present invention provides an OCT image layering and lesion semantic segmentation apparatus, comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is configured to operate in accordance with the instructions to perform the steps of the method according to any of the above.
In a fourth aspect, the invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of any of the methods described above.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides an OCT image layering and focus semantic segmentation method, device and storage medium, because the training data of the second stage network is concentrated in the 'concerned area', the learning of the model is more specific to the main affected area of the age-related macular degeneration focus; because the second stage network reduces the sample input of irrelevant areas, the proportion of focus pixels in the total pixels is improved through phase change, and the learning difficulty of small-area positive focuses such as drusen is reduced.
Drawings
Fig. 1 is a flowchart of an OCT image layering and lesion semantic segmentation method according to an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
Example 1
The embodiment introduces an OCT image layering and lesion semantic segmentation method, an OCT image layering and lesion semantic segmentation device, and a storage medium, including:
acquiring image data of each BScan two-dimensional slice of OCT of one eye of a patient to be inspected, and performing normalization processing;
inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layering, the optic disc region and the macular depression region in the first stage;
according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the macular depression area in the first stage, a 'concerned area' is manufactured;
acquiring a data map of the attention area, and carrying out normalization processing on the data;
and inputting the normalized data image of the attention area into the convolution neural network of the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus of the second stage.
The application process of the OCT image layering and lesion semantic segmentation method, device, and storage medium provided in this embodiment specifically involves the following steps:
first, training, verifying network stage
1. Firstly, amplifying and normalizing data of BScan two-dimensional slice image data of OCT and semantic segmentation labels of corresponding layering, optic disc areas and yellow spot sunken areas, and randomly disordering to manufacture a data set of a first stage;
2. at each step of the first-stage network training, inputting image data processed in a data set at the first stage into a convolutional neural network at the first stage to obtain a semantic segmentation prediction result of a layering layer, a optic disc region and a macular depression region at the first stage, calculating loss with a corresponding label, and performing gradient back propagation and network parameter updating, wherein an optimizer of the network adopts Adam;
3. the "region of interest" is created from the BScan two-dimensional slice image data of OCT and the semantic segmentation labels of the corresponding layers, optic disc region, and macular dip region. The specific mode is as follows:
(1) selecting a slice with the thinnest thickness of a nerve fiber layer where the central point of a yellow spot depressed area is located in a data label of the same-eye OCT, wherein the central point of the yellow spot depressed area is used as the central point of the yellow spot fovea of the OCT image of the eye;
(2) selecting the maximum value of the width of an optic disc area in a data label of the same-eye OCT as the 'optic disc diameter' of the OCT image;
(3) according to the horizontal and longitudinal physical resolution and pixel resolution values under the specific mode of the OCT model, the horizontal pixel number of 2 optic disc diameters is converted into the horizontal physical size, the central concave point of the macula lutea is taken as the center of a circle, the physical value of the optic disc diameter is taken as the radius to determine a 'attention area circle', and the attention area circle is converted into the pixel width of an attention area corresponding to the attention area circle in each slice
(4) Taking the column of the central foveal point of the macula lutea as a central point on the slice, obtaining the pixel width through the conversion in the step (3), taking the center of a polygonal external rectangle of the fundus layer structure label non-background areas of the columns as the center, and taking the specific pixel number as the height (the height is an empirical value set according to the statistic value of the layer height of the OCT image), and obtaining a 'concerned area';
4. taking a data graph and a layering graph of a focus area, drusen, pigment epithelium layer separation and a focus label graph of choroidal neovascularization for data augmentation, normalizing the data, and randomly disordering to prepare a data set of a second stage;
5. in each step of the second stage of network training, image data processed in a second stage of data set is input into the convolutional neural network in the first stage to obtain the semantic segmentation prediction results of the layering, drusen, pigment epithelium layer separation and choroidal neovascular lesions in the second stage, loss calculation is carried out on the semantic segmentation prediction results and corresponding labels, gradient back propagation and network parameter updating are carried out, and an optimizer of the network adopts Adam;
second, model reasoning phase
1. Normalizing image data of each BScan two-dimensional slice of the OCT of one eye of a patient to be detected;
2. inputting the slice image data into a trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layering, the optic disc region and the macular pucker region in the first stage;
3. based on the obtained hierarchical layer of the BScan two-dimensional slice image data of the eye OCT, the semantic segmentation prediction results of the optic disc region and the macular dip region, a "region of interest" is created. The specific mode is similar to the mode of making a 'key area' in the training and verification stages:
(1) selecting a slice with the thinnest thickness of a nerve fiber layer in a row where the central point of a yellow spot depressed area of a first-stage network prediction result of same-eye OCT is located, wherein the central point of the yellow spot depressed area is used as the central point of the central depression of the yellow spot of an OCT image of the eye; if no macular pucker area is predicted on all slices, adopting the scheme in (3);
(2) selecting the maximum value of the width of the optic disc area in the prediction results of all the slices of the network in the first stage of the same-eye OCT as the optic disc diameter of the OCT image; if no optic disc area is predicted on all the slices, adopting the scheme in the step (3);
(3) when the macular pucker area is not predicted on all the slices or the optic disc area is not predicted on all the slices, taking a complete image of the whole slice of the 'attention area' of each slice;
(4) according to the horizontal and longitudinal physical resolution and pixel resolution values in the specific mode of the OCT model, converting the horizontal pixel number of 2 optic disc diameters into the horizontal physical size, determining a 'attention area circle' by taking the central concave point of the yellow spot as the center of a circle and the physical value of the optic disc diameter as the radius, and converting the horizontal pixel number into the pixel width of an 'attention area' corresponding to the 'attention area circle' in each slice;
(4) taking the column of the central foveal point of the macula lutea as a central point on the slice, obtaining the pixel width through the conversion in the step (3), taking the center of a polygonal circumscribed rectangle of a non-background area in the fundus layer structure prediction results of the columns as the center, and taking a specific pixel number as the height (the height is an empirical value set according to the layer height statistic value of the OCT image), and obtaining a 'concerned area';
4. taking a data map of the attention area to normalize the data;
5. and inputting the normalized data image of the attention area into the convolution neural network of the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus of the second stage.
Example 2
The present embodiment provides an OCT image layering and lesion semantic segmentation apparatus, including:
the system comprises a first acquisition unit, a second acquisition unit and a third acquisition unit, wherein the first acquisition unit is used for acquiring each BScan two-dimensional slice image data of the OCT of one eye of a patient to be inspected and carrying out normalization processing;
the first input unit is used for inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layers, the optic disc region and the macular depression region in the first stage;
the processing unit is used for manufacturing a focus area according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the yellow spot sunken area in the first stage;
the second acquisition unit is used for acquiring a data map of the attention area and normalizing the data;
and the second input unit is used for inputting the normalized data image of the attention area into the convolution neural network at the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus at the second stage.
Example 3
The embodiment provides an OCT image layering and focus semantic segmentation device, which comprises a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is configured to operate in accordance with the instructions to perform the steps of the method according to any of embodiment 1.
Example 4
The present embodiment provides a computer-readable storage medium, on which a computer program is stored, which program, when being executed by a processor, carries out the steps of the method of any of the embodiment 1.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (8)

1. An OCT image layering and focus semantic segmentation method is characterized by comprising the following steps:
acquiring image data of each BScan two-dimensional slice of OCT of one eye of a patient to be inspected, and performing normalization processing;
inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layering, the optic disc region and the macular depression region in the first stage;
according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the macular depression area in the first stage, a 'concerned area' is manufactured;
acquiring a data map of the attention area, and carrying out normalization processing on the data;
and inputting the normalized data image of the attention area into the convolution neural network of the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus of the second stage.
2. The OCT image layering and lesion semantic segmentation method of claim 1, wherein: creating a "region of interest" based on the semantic segmentation prediction results of the first stage of the hierarchy, the optic disc region, and the macular pucker region, comprising:
selecting a slice with the thinnest thickness of a nerve fiber layer where the central point of a yellow spot depressed area is located in a data label of the same-eye OCT, wherein the central point of the yellow spot depressed area is used as the central point of the yellow spot fovea of the OCT image of the eye;
selecting the maximum value of the width of the optic disc area in all the section data labels of the same-eye OCT as the 'optic disc diameter' of the OCT image;
according to the horizontal and longitudinal physical resolution and pixel resolution values in the specific mode of the OCT model, converting the horizontal pixel number of 2 optic disc diameters into the horizontal physical size, determining a 'attention area circle' by taking the central concave point of the yellow spot as the center of a circle and the physical value of the optic disc diameter as the radius, and converting the horizontal pixel number into the pixel width of an 'attention area' corresponding to the 'attention area circle' in each slice;
and taking the column of the central foveal point of the macula as a central point on the section, obtaining the pixel width through conversion of the width, taking the center of a polygonal external rectangle of the fundus layer structure label non-background areas of the columns as the center, and taking the specific pixel number as the height, and obtaining the 'region of interest'.
3. The OCT image layering and lesion semantic segmentation method of claim 1, wherein: the training method of the convolutional neural network of the first stage comprises the following steps:
acquiring BScan two-dimensional slice image data of OCT and corresponding semantic segmentation labels of layering, optic disc area and yellow spot sunken area, performing data augmentation and normalization, and randomly disordering to manufacture a data set of a first stage;
at each step of the first-stage network training, inputting image data processed in a data set at the first stage into a convolutional neural network at the first stage to obtain a semantic segmentation prediction result of the layering, the optic disc region and the macular depression region at the first stage, calculating loss with a corresponding label, and performing gradient back propagation and network parameter updating;
according to BScan two-dimensional slice image data of OCT and corresponding semantic segmentation labels of layering, optic disc area and yellow spot sunken area, making a 'key area';
acquiring a data map and a layered focus label map of a 'key region', drusen, pigment epithelial layer detachment and choroidal neovascularization, performing data amplification, normalizing and randomly disordering the data, and making into a data set of a second stage;
and in each step of the second stage of network training, inputting the image data processed in the second stage of data set into the convolutional neural network in the first stage to obtain the semantic segmentation prediction results of the layering, drusen, pigment epithelium layer separation and choroidal neovascular lesion in the second stage, calculating loss with the corresponding label, and performing gradient back propagation and network parameter updating.
4. The OCT image layering and lesion semantic segmentation method of claim 3, wherein: the method for making the key area is the same as the method for making the attention area.
5. The OCT image layering and lesion semantic segmentation method of claim 3, wherein: further comprising: and optimizing the network by adopting an Adam optimizer.
6. An OCT image layering and lesion semantic segmentation device is characterized by comprising:
the system comprises a first acquisition unit, a second acquisition unit and a third acquisition unit, wherein the first acquisition unit is used for acquiring each BScan two-dimensional slice image data of the OCT of one eye of a patient to be inspected and carrying out normalization processing;
the first input unit is used for inputting the normalized slice image data into a pre-trained convolutional neural network in a first stage one by one to obtain semantic segmentation prediction results of the layers, the optic disc region and the macular depression region in the first stage;
the processing unit is used for manufacturing a focus area according to the obtained semantic segmentation prediction results of the layering, the optic disc area and the yellow spot sunken area in the first stage;
the second acquisition unit is used for acquiring a data map of the attention area and normalizing the data;
and the second input unit is used for inputting the normalized data image of the attention area into the convolution neural network at the second stage to obtain semantic segmentation prediction results of layering, drusen, pigment epithelial layer separation and choroidal neovascularization focus at the second stage.
7. An OCT image layering and focus semantic segmentation device is characterized in that: comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is configured to operate in accordance with the instructions to perform the steps of the method according to any one of claims 1 to 5.
8. A computer-readable storage medium having stored thereon a computer program, characterized in that: the program when executed by a processor implements the steps of the method of any one of claims 1 to 5.
CN202111675172.5A 2021-12-31 2021-12-31 OCT image layering and focus semantic segmentation method, device and storage medium Pending CN114359219A (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202111675172.5A CN114359219A (en) 2021-12-31 2021-12-31 OCT image layering and focus semantic segmentation method, device and storage medium
PCT/CN2022/088759 WO2023123765A1 (en) 2021-12-31 2022-04-24 Oct image layering and lesion semantic segmentation method and apparatus, and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111675172.5A CN114359219A (en) 2021-12-31 2021-12-31 OCT image layering and focus semantic segmentation method, device and storage medium

Publications (1)

Publication Number Publication Date
CN114359219A true CN114359219A (en) 2022-04-15

Family

ID=81105685

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111675172.5A Pending CN114359219A (en) 2021-12-31 2021-12-31 OCT image layering and focus semantic segmentation method, device and storage medium

Country Status (2)

Country Link
CN (1) CN114359219A (en)
WO (1) WO2023123765A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023123765A1 (en) * 2021-12-31 2023-07-06 苏州比格威医疗科技有限公司 Oct image layering and lesion semantic segmentation method and apparatus, and storage medium

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118096769A (en) * 2024-04-29 2024-05-28 中国科学院宁波材料技术与工程研究所 Retina OCT image analysis method and device

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020056454A1 (en) * 2018-09-18 2020-03-26 MacuJect Pty Ltd A method and system for analysing images of a retina
WO2020125806A1 (en) * 2018-12-17 2020-06-25 Shanghai United Imaging Intelligence Co., Ltd. Systems and methods for image segmentation
CN109886946B (en) * 2019-02-18 2023-05-23 广州视源电子科技股份有限公司 Deep learning-based early senile maculopathy weakening supervision and classification method
CN110148111A (en) * 2019-04-01 2019-08-20 江西比格威医疗科技有限公司 The automatic testing method of a variety of retina lesions in a kind of retina OCT image
CN114359219A (en) * 2021-12-31 2022-04-15 苏州比格威医疗科技有限公司 OCT image layering and focus semantic segmentation method, device and storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023123765A1 (en) * 2021-12-31 2023-07-06 苏州比格威医疗科技有限公司 Oct image layering and lesion semantic segmentation method and apparatus, and storage medium

Also Published As

Publication number Publication date
WO2023123765A1 (en) 2023-07-06

Similar Documents

Publication Publication Date Title
Imran et al. Comparative analysis of vessel segmentation techniques in retinal images
Almotiri et al. Retinal vessels segmentation techniques and algorithms: a survey
Odstrcilik et al. Retinal vessel segmentation by improved matched filtering: evaluation on a new high‐resolution fundus image database
Voets et al. Replication study: Development and validation of deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs
El‐Hag et al. Classification of retinal images based on convolutional neural network
Hassan et al. RAG-FW: A hybrid convolutional framework for the automated extraction of retinal lesions and lesion-influenced grading of human retinal pathology
CN112132817A (en) Retina blood vessel segmentation method for fundus image based on mixed attention mechanism
Hassan et al. Joint segmentation and quantification of chorioretinal biomarkers in optical coherence tomography scans: A deep learning approach
CN114359219A (en) OCT image layering and focus semantic segmentation method, device and storage medium
CN103348359A (en) 3D retinal disruptions detection using optical coherence tomography
CN108305241B (en) SD-OCT image GA lesion segmentation method based on depth voting model
CN108416793B (en) Choroidal vessel segmentation method and system based on three-dimensional coherence tomography image
CN111862009A (en) Classification method of fundus OCT images and computer-readable storage medium
Ali et al. Glaucoma detection based on local binary patterns in fundus photographs
CN113160226A (en) Two-way guide network-based classification segmentation method and system for AMD lesion OCT image
CN115004222A (en) Neural network processing of OCT data to generate predictions of geographic atrophy growth rate
Hassan et al. Multilayered deep structure tensor delaunay triangulation and morphing based automated diagnosis and 3D presentation of human macula
Javidi et al. Retinal image assessment using bi-level adaptive morphological component analysis
WO2023036899A1 (en) Method and system for retinal tomography image quality control
Kadan et al. Optimized hybrid classifier for diagnosing diabetic retinopathy: iterative blood vessel segmentation process
Yadav et al. Automatic Cataract Severity Detection and Grading Using Deep Learning
Mookiah et al. Computer aided diagnosis of diabetic retinopathy using multi-resolution analysis and feature ranking frame work
CN117577319A (en) Daily monitoring and early warning method for diabetic retinopathy
CN113592843B (en) Improved U-Net fundus retina blood vessel image segmentation method and device
Li et al. Automatic quantification of cone photoreceptors in adaptive optics scanning light ophthalmoscope images using multi-task 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