CN106846338A - Retina OCT image based on mixed model regards nipple Structural Techniques - Google Patents
Retina OCT image based on mixed model regards nipple Structural Techniques Download PDFInfo
- Publication number
- CN106846338A CN106846338A CN201710071069.7A CN201710071069A CN106846338A CN 106846338 A CN106846338 A CN 106846338A CN 201710071069 A CN201710071069 A CN 201710071069A CN 106846338 A CN106846338 A CN 106846338A
- Authority
- CN
- China
- Prior art keywords
- image
- segmentation
- image based
- oct image
- shape
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 210000001525 retina Anatomy 0.000 title claims abstract description 18
- 210000002445 nipple Anatomy 0.000 title claims abstract description 14
- 230000011218 segmentation Effects 0.000 claims abstract description 28
- 230000004256 retinal image Effects 0.000 claims abstract description 22
- 210000003733 optic disk Anatomy 0.000 claims abstract description 17
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 239000011159 matrix material Substances 0.000 claims description 5
- 239000012528 membrane Substances 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 241001269238 Data Species 0.000 claims description 3
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 239000004576 sand Substances 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 2
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 238000004364 calculation method Methods 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 238000005192 partition Methods 0.000 claims 1
- 238000003384 imaging method Methods 0.000 abstract description 8
- 230000003287 optical effect Effects 0.000 abstract description 3
- 238000012014 optical coherence tomography Methods 0.000 description 16
- 238000004458 analytical method Methods 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000005252 bulbus oculi Anatomy 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000001508 eye Anatomy 0.000 description 1
- 210000003128 head Anatomy 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000002207 retinal effect Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Eye Examination Apparatus (AREA)
- Image Analysis (AREA)
Abstract
Nipple Structural Techniques are regarded the invention discloses a kind of retina OCT image based on mixed model, is comprised the following steps:1)Retinal images are filtered, even up treatment by image preprocessing;2)According to the mark point of hand labeled, active appearance models are set up, and the coarse segmentation of retinal images structure is carried out with active appearance models;3)With step 2)In result be constraints, the Accurate Segmentation of retinal images structure is carried out with graph search method.The present invention provide first it is a kind of with feasibility and validity to the SD OCT centered on optic disk(Domain optical coherence fault imaging)The method that retinal images are split, makes microscope imaging equipment that the imaging of noncontact, high-resolution, high-res can be in this way carried out with profit.
Description
Technical field
The invention belongs to image processing method, especially to SD-OCT (the domain optical coherence tomographies centered on optic disk
Imaging) retinal images dividing method.
Background technology
Optical coherence tomography (optical coherence tomography, OCT) is developed rapidly in recent years
A kind of imaging technique, and be gradually widely used in gathering high-resolution retinal images.The retinal structure of human eye is very multiple
Miscellaneous, especially with the region centered on nipple (Optical Nerve Head, ONH), its structure change is increasingly complex, change
It is more notable, so, develop a kind of retina of reliable automation and regard nipple area segmentation of structures and measuring method just very
It is important.
So far, it has been disclosed that some researchs mainly retina OCT images of macular area are done with the segmentation of each layer of retina,
For the segmentation of the retina OCT image in the region centered on regarding nipple, although there are some researchs to be directed to this region and done
Segmentation of structures, but it is to split optic disk, optic cup based on the colored fundus photograph of two dimension to have certain methods;Although have certain methods is
Split based on three-dimensional OCT image, but because the complexity of ONH structures, and the structure change for being influenceed to bring by lesion,
So as to the precision for causing segmentation is not very high.
The content of the invention
The technical problems to be solved by the invention are the defects for overcoming prior art, there is provided one kind is with feasibility and effectively
The dividing method to SD-OCT (domain optical coherence fault imaging) retinal images centered on optic disk of property.
In order to solve the above technical problems, the present invention provides a kind of retina OCT image based on mixed model regards nipple knot
Structure dividing method.
It is automatic based on AAM (active appearance models) and Graph-search (graph search) technology the invention provides one kind
Dividing method, the method mainly includes 3 steps:Image preprocessing, AAM modelings and coarse segmentation, Graph-search (search by figure
Rope) method essence segmentation.
1) retinal images are filtered, even up treatment by image preprocessing;
2) according to the mark point of hand labeled, AAM (active appearance models) is set up, and it is complete with AAM (active appearance models)
Into the coarse segmentation of retinal images structure;
3) with step 2) in result be constraints, realize retinal images with Graph-Search (graph search) method
The essence segmentation of structure.
Step 1) in, image filtering treatment removes noise using the algorithm of gradient anisotropy parameter.
Step 1) in, image evens up treatment by the use of external limiting membrane as the benchmark evened up, the extraneous film location that will first detect
A fixed value is appointed as, then on the basis of this fixed value, align full figure.
Step 2) in, it is the step of set up active appearance models:
First through step 1) contour line for marking segmentation object on each two field picture for obtaining manually is pre-processed, then will
The segmentation object of 3-dimensional is expressed as the storehouse of the contour line of 2 dimensions, and multiple mark points are marked on the contour line of mark;
After the mark point of all training datas all marks completion, set up using the active appearance models of standard and regarded
The model of nethike embrane picture structure.
Active appearance models include two parts of shape and texture model:
Wherein,It is average shape model,It is the average texture model corresponding to this average shape, QsAnd QgIt is by master
Shape, the transformation matrix of texture principal component characteristic component formation that constituent analysis is calculated.S is the shape for controlling change in shape
Parameter;T is the parametric texture for controlling texture variations;X is shape, and g is texture model;Using active appearance models in test
Positioned in data, split area-of-interest, obtain layer 1- layers 7 of initial results.
Step 3) in, use is comprised the following steps that:
First, retinal images are defined as three-dimensional matrice I (x, y, z), its size is X × Y × Z, wherein x, y, z is empty
Between coordinate, X, Y, Z are respectively the voxel numbers on three directions;
Surface to be detected is defined as function S (x, y), x ∈ { 0 ..., X-1 }, y ∈ { 0 ..., Y-1 }, and S (x, y) ∈
{ 0 ..., Z-1 };
Parameter, Δ x defines the smoothness constraint condition on x directions, and parameter, Δ y defines the smoothness constraint condition on y directions;Make
Ultimate range of the two neighboring surface in x directions meets | and S (x+1, y)-S (x, y) |≤Δ x, the ultimate range in y directions meets
| S (x, y+1)-S (x, y) |≤Δ y;
Then, a summit-weight digraph G=(V, E) is set up based on image voxel, it includes a collection for vertex v
Close V and set E of side e;In this digraph, any vertex v ∈ V correspond to the individuality in image I (x, y, z)
Vegetarian refreshments, and any one arc<vi, vj>∈ E are connected to two vertex vsi、vj;Each vertex v ∈ V, V (x, y, z) | (z > 0) }
Cost value c (x, y, z) calculated according to the gradient magnitude of OCT image, indicate a pixel and be not belonging to target surface
Possibility, the weights of each vertex v ∈ V calculate according to cost value, such as following formula (2):
The problem for searching optimal surface is changed into the retrieval minimum cost closed set in digraph G, the sense in image is emerging
Interesting region segmentation is out.
The step of also including to optic disk image segmentation:
The border of optic disk image is detected, and layer information is concealed in the part in optic disk region;Obtained according to layering result
The projected image in the z directions between layer 6 and layer 7, recycles shape prior model algorithm to be partitioned into optic disk area image.
The beneficial effect that the present invention is reached:
The present invention provide first it is a kind of with feasibility and validity to SD-OCT (the frequency domain light centered on optic disk
Learn coherence tomography) method split of retinal images, microscope imaging equipment is in this way carried out with profit
Noncontact, high-resolution, the imaging of high-res.
Brief description of the drawings
Fig. 1 is the flow chart of the retina OCT image ONH segmentation of structures algorithms based on mixed model.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following examples are only used for clearly illustrating the present invention
Technical scheme, and can not be limited the scope of the invention with this.
The fundamental block diagram of the inventive method as shown in Figure 1, mainly includes 3 steps:Image preprocessing, with AAM (actively
Display model algorithm) initial segmentation is done, with Graph-Search (graph search algorithm does fine segmentation).It is described in detail below.
(1) image preprocessing
Image preprocessing is mainly included the following steps that:The filtering of retinal images and even up.
The filtering of (a) image
Treatment is filtered to image using the algorithm of gradient anisotropy parameter, noise is eliminated as much as and is retained image
Detailed information.
B () image is evened up
During retina OCT image is gathered, because the Rotation of eyeball of subject, cause the difference of the image of collection
Larger saltus step is had between frame, it is not high enough in the smoothness of some location drawing pictures, suffer from this, cause layering result to be deteriorated.
Operation is evened up to image, is by the use of being easier the external limiting membrane of detection as the benchmark evened up, the external limiting membrane that will first detect
Position is appointed as a fixed value, and then as benchmark, align full figure.
(2) AAM models (active appearance models) are set up
The contour line of segmentation object is marked manually on each two field picture of collection first, then by the segmentation object of 3-dimensional
It is expressed as the storehouse of the contour line of 2 dimensions.Multiple mark points are marked on the contour line of mark, is labelled with altogether in the present embodiment
13 contour lines, except an ILM (internal limiting membrane) is labelled with 16 mark points, remaining 12 profile is all only labelled with 6 marks
Point.
After the mark point of all training datas all marks completion, the AAM (active appearance models) using standard comes
Set up the model of retinal images structure.
AAM models (active appearance models) include two parts of shape s and texture model g, as shown in formula (1):
Wherein,It is average shape model,It is the average texture model corresponding to this average shape, QsAnd QgIt is by master
Shape, the transformation matrix of texture principal component characteristic component formation that constituent analysis is calculated.S is the shape for controlling change in shape
Parameter;T is the parametric texture for controlling texture variations.X is according to average shape modelThe shape obtained with transformation matrix and parameter
Model, g is according to average texture modelThe texture model obtained with transformation matrix and parameter, just can be using this model
Positioned in test data, split area-of-interest, obtain layer 1- layers 7 of initial results.
(3) Accurate Segmentation of retinal images structure is realized with Graph-Search (graph search) method
The segmentation result obtained with AAM models (active appearance models) is accurate not enough, thus the present invention propose with
AAM models (active appearance models) result is constraints, and retinal images are realized with Graph-Search (graph search) method
The method of structure Accurate Segmentation.
Single face detection 3-D Graph-Search (graph search) method that K.li et al. is proposed defines retinal images
Into three-dimensional matrice I (x, y, z), its size is X × Y × Z, wherein, x, y, z is space coordinates, and X, Y, Z are respectively on three directions
Voxel number.Surface to be detected can be defined as function S (x, y), x ∈ { 0 ..., X-1 }, y ∈ { 0 ..., y-1 }, and S
(x, y) ∈ { 0 ..., Z-1 }.In addition, parameter, Δ x defines the smoothness constraint condition on x directions, parameter, Δ y defines y directions
On smoothness constraint condition.That is ultimate range of the two neighboring surface in x directions will meet | S (x+1, y)-S (x, y) |
≤ Δ x, the ultimate range in y directions will meet | S (x, y+1)-S (x, y) |≤Δ y.Cost value c (x, y, z) table of each point
Understand that a voxel is not belonging to the possibility of target surface.So, just there is the cost value of minimum on optimal surface.Then, it is based on
Image voxel sets up a summit-weight digraph G=(V, E), its set V for including vertex v and a set of side e
E.In this digraph, any vertex v ∈ V correspond to a tissue points in image I (x, y, z), and any one arc<
vi, vj>∈ E are connected to two vertex vsi、vj.Each vertex v ∈ V, the cost value c (x, y, z) of { V (x, y, z) | (z > 0) } are
What the gradient magnitude according to OCT image was calculated, the possibility that a pixel is not belonging to target surface is indicated, each top
The weights of point v ∈ V are calculated according to cost value again, such as following formula (2):
So, the problem for searching optimal surface has been converted to retrieve minimum cost closed set in digraph G, such that it is able to
Minimal closed set is calculated using max-flow/minimal cut algorithm, the area-of-interest in image is split.
In this algorithm, cost value calculates the gradient magnitude in z directions using Sobel boundary operators.It is respectively adopted
Two kinds of cost equation, wherein, layer 1,5,6 is from top to bottom, by black to white change;Layer 2,3,4,7 be from top to bottom,
By the change of white to black.
(4) segmentation of optic disk image
Inside nerve channel image, the hierarchy of retinal images tissue does not know that the method for AUTOMATIC ZONING cannot
Effectively layering, therefore, the border of optic disk image is detected in this algorithm, and layer information is concealed in the part in optic disk region.It is first
The projected image in the z directions between layer 6 and layer 7 is first obtained according to layering result.Shape prior model algorithm is recycled to be partitioned into
Optic disk region.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, on the premise of the technology of the present invention principle is not departed from, some improvement and deformation can also be made, these improve and deform
Also should be regarded as protection scope of the present invention.
Claims (7)
1. a kind of retina OCT image based on mixed model regards nipple Structural Techniques, it is characterized in that, including following step
Suddenly:
1) retinal images are filtered, even up treatment by image preprocessing;
2) according to the mark point of hand labeled, active appearance models are set up, and retinal images knot is carried out with active appearance models
The coarse segmentation of structure;
3) with step 2) in result be constraints, the Accurate Segmentation of retinal images structure is carried out with graph search method.
2. the retina OCT image based on mixed model according to claim 1 regards nipple Structural Techniques, its feature
It is, step 1) in, image filtering treatment removes noise using the algorithm of gradient anisotropy parameter.
3. the retina OCT image based on mixed model according to claim 1 regards nipple Structural Techniques, its feature
It is, step 1) in, image evens up treatment by the use of external limiting membrane as the benchmark evened up, and the extraneous film location that will first detect is appointed as
One fixed value, then on the basis of this fixed value, align full figure.
4. the retina OCT image based on mixed model according to claim 1 regards nipple Structural Techniques, its feature
It is, step 2) in, it is the step of set up active appearance models:
First through step 1) contour line for marking segmentation object on each two field picture for obtaining manually is pre-processed, then by 3-dimensional
Segmentation object be expressed as 2 dimensions contour line storehouse, multiple mark points are marked on the contour line of mark;
After the mark point of all training datas all marks completion, retina is set up using the active appearance models of standard
The model of picture structure.
5. the retina OCT image based on mixed model according to claim 4 regards nipple Structural Techniques, its feature
It is that active appearance models include two parts of shape and texture model:
Wherein,It is average shape model,It is the average texture model corresponding to this average shape, QsAnd QgIt is by principal component
Shape, the transformation matrix of texture principal component characteristic component formation that analytical calculation is obtained.S is the shape ginseng for controlling change in shape
Number;T is the parametric texture for controlling texture variations;X is shape, and g is texture model;Using active appearance models in test number
According to upper positioning, segmentation area-of-interest, layer 1- layers 7 of initial results are obtained.
6. the retina OCT image based on mixed model according to claim 1 regards nipple Structural Techniques, its feature
It is, step 3) in, use is comprised the following steps that:
First, retinal images are defined as three-dimensional matrice I (x, y, z), its size is X × Y × Z, wherein x, y, z is that space is sat
Mark, X, Y, Z are respectively the voxel numbers on three directions;
Surface to be detected is defined as function S (x, y), x ∈ { 0 ..., X-1 }, y ∈ { 0 ..., Y-1 }, and S (x, y) ∈
{ 0 ..., Z-1 };
Parameter, Δ x defines the smoothness constraint condition on x directions, and parameter, Δ y defines the smoothness constraint condition on y directions;Make adjacent
Ultimate range of two surfaces in x directions meets | and S (x+1, y)-S (x, y) |≤Δ x, the ultimate range in y directions meets | S
(x, y+1)-S (x, y) |≤Δ y;
Then, a summit-weight digraph G=(V, E) is set up based on image voxel, its include a set V for vertex v and
One set E of side e;In this digraph, any vertex v ∈ V correspond to a tissue points in image I (x, y, z),
And any one arc<vi, vj>∈ E are connected to two vertex vsi、vj;Each vertex v ∈ V, the cost of { V (x, y, z) | (z > 0) }
Value c (x, y, z) is calculated according to the gradient magnitude of OCT image, indicates the possibility that a pixel is not belonging to target surface
Property, the weights of each vertex v ∈ V are according to cost value calculating, such as following formula (2):
The problem for searching optimal surface is changed into the retrieval minimum cost closed set in digraph G, by the region of interest in image
Regional partition is out.
7. the retina OCT image based on mixed model according to claim 5 regards nipple Structural Techniques, its feature
Be, also including to optic disk image segmentation the step of:
The border of optic disk image is detected, and layer information is concealed in the part in optic disk region;The He of layer 6 is obtained according to layering result
The projected image in the z directions between layer 7, recycles shape prior model algorithm to be partitioned into optic disk area image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710071069.7A CN106846338A (en) | 2017-02-09 | 2017-02-09 | Retina OCT image based on mixed model regards nipple Structural Techniques |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710071069.7A CN106846338A (en) | 2017-02-09 | 2017-02-09 | Retina OCT image based on mixed model regards nipple Structural Techniques |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106846338A true CN106846338A (en) | 2017-06-13 |
Family
ID=59122409
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710071069.7A Pending CN106846338A (en) | 2017-02-09 | 2017-02-09 | Retina OCT image based on mixed model regards nipple Structural Techniques |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106846338A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107845088A (en) * | 2017-10-25 | 2018-03-27 | 苏州比格威医疗科技有限公司 | Physiological parameter acquisition algorithm in retina OCT image based on dynamic constrained graph search |
CN109886965A (en) * | 2019-04-09 | 2019-06-14 | 山东师范大学 | The layer of retina dividing method and system that a kind of level set and deep learning combine |
CN109886969A (en) * | 2019-01-15 | 2019-06-14 | 南方医科大学 | A kind of three-dimensional automatic division method of respiratory tract endoscopic optical coherent faultage image |
CN109993726A (en) * | 2019-02-21 | 2019-07-09 | 上海联影智能医疗科技有限公司 | Detection method, device, equipment and the storage medium of medical image |
CN113724203A (en) * | 2021-08-03 | 2021-11-30 | 唯智医疗科技(佛山)有限公司 | Segmentation method and device for target features in OCT (optical coherence tomography) image |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1926573A (en) * | 2004-01-30 | 2007-03-07 | 思代软件公司 | System and method for applying active appearance models to image analysis |
CN101369309A (en) * | 2008-09-26 | 2009-02-18 | 北京科技大学 | Human ear image normalization method based on active apparent model and outer ear long axis |
-
2017
- 2017-02-09 CN CN201710071069.7A patent/CN106846338A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1926573A (en) * | 2004-01-30 | 2007-03-07 | 思代软件公司 | System and method for applying active appearance models to image analysis |
CN101369309A (en) * | 2008-09-26 | 2009-02-18 | 北京科技大学 | Human ear image normalization method based on active apparent model and outer ear long axis |
Non-Patent Citations (3)
Title |
---|
KANG LI ET AL: ""Optimal Surface Segmentation in Volumetric Images-A Graph-Theoretic Approach"", 《IEEE》 * |
蔡凡: ""基于主动外观模型的图像分割研究"", 《闽江学院学报》 * |
陆圣陶: ""基于三维图搜索的SDOCT视网膜图像层边界分割与研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107845088A (en) * | 2017-10-25 | 2018-03-27 | 苏州比格威医疗科技有限公司 | Physiological parameter acquisition algorithm in retina OCT image based on dynamic constrained graph search |
WO2019080215A1 (en) * | 2017-10-25 | 2019-05-02 | 苏州比格威医疗科技有限公司 | Algorithm for acquiring physiological parameters in retinal oct image based on dynamic constraint graph search |
CN107845088B (en) * | 2017-10-25 | 2020-02-07 | 苏州比格威医疗科技有限公司 | Method for acquiring physiological parameters in retina OCT image based on dynamic constraint graph search |
CN109886969A (en) * | 2019-01-15 | 2019-06-14 | 南方医科大学 | A kind of three-dimensional automatic division method of respiratory tract endoscopic optical coherent faultage image |
CN109993726A (en) * | 2019-02-21 | 2019-07-09 | 上海联影智能医疗科技有限公司 | Detection method, device, equipment and the storage medium of medical image |
CN109886965A (en) * | 2019-04-09 | 2019-06-14 | 山东师范大学 | The layer of retina dividing method and system that a kind of level set and deep learning combine |
CN113724203A (en) * | 2021-08-03 | 2021-11-30 | 唯智医疗科技(佛山)有限公司 | Segmentation method and device for target features in OCT (optical coherence tomography) image |
CN113724203B (en) * | 2021-08-03 | 2024-04-23 | 唯智医疗科技(佛山)有限公司 | Model training method and device applied to target feature segmentation in OCT image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106846338A (en) | Retina OCT image based on mixed model regards nipple Structural Techniques | |
EP3671324B1 (en) | Method, device and computer program for virtual adapting of a spectacle frame | |
CN107358648B (en) | Real-time full-automatic high quality three-dimensional facial reconstruction method based on individual facial image | |
Miri et al. | Multimodal segmentation of optic disc and cup from SD-OCT and color fundus photographs using a machine-learning graph-based approach | |
CN106997605B (en) | A method of foot type video is acquired by smart phone and sensing data obtains three-dimensional foot type | |
US7992999B2 (en) | Automated assessment of optic nerve head with spectral domain optical coherence tomography | |
Bérard et al. | High-quality capture of eyes. | |
CN110033465A (en) | A kind of real-time three-dimensional method for reconstructing applied to binocular endoscope medical image | |
CN107240129A (en) | Object and indoor small scene based on RGB D camera datas recover and modeling method | |
CN104408462B (en) | Face feature point method for rapidly positioning | |
CN106037931A (en) | Method and system for advanced transcatheter aortic valve implantation planning | |
CN102903135A (en) | Method and apparatus for realistic simulation of wrinkle aging and de-aging | |
CN110232389A (en) | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance | |
CN105913013A (en) | Binocular vision face recognition algorithm | |
CN107610202A (en) | Marketing method, equipment and the storage medium replaced based on facial image | |
CN106408576B (en) | Automatic region of interest segmentation method and system based on three-dimensional ultrasonic image | |
CN105894508B (en) | A kind of medical image is automatically positioned the appraisal procedure of quality | |
CN103247056B (en) | Human bone articular system three-dimensional model-bidimensional image spatial registration method | |
CN110458752A (en) | A kind of image based under the conditions of partial occlusion is changed face method | |
CN108053398A (en) | A kind of melanoma automatic testing method of semi-supervised feature learning | |
WO2016026570A1 (en) | Determining user data based on image data of a selected eyeglass frame | |
CN102567734A (en) | Specific value based retina thin blood vessel segmentation method | |
WO2020252969A1 (en) | Eye key point labeling method and apparatus, and training method and apparatus for eye key point detection model | |
CN104318565B (en) | Interactive method for retinal vessel segmentation based on bidirectional region growing of constant-gradient distance | |
CN114694236A (en) | Eyeball motion segmentation positioning method based on cyclic residual convolution neural network |
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: 20170613 |
|
RJ01 | Rejection of invention patent application after publication |