CN109754081A - A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning - Google Patents
A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning Download PDFInfo
- Publication number
- CN109754081A CN109754081A CN201811599884.1A CN201811599884A CN109754081A CN 109754081 A CN109754081 A CN 109754081A CN 201811599884 A CN201811599884 A CN 201811599884A CN 109754081 A CN109754081 A CN 109754081A
- Authority
- CN
- China
- Prior art keywords
- image
- module
- hysteroscope
- data
- deep learning
- 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.)
- Withdrawn
Links
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to medical domains, specifically disclose a kind of hysteroscope medical image identifying system based on deep learning, it include: alimentary canal hysteroscope module, autonomous database, image segmentation module, data computation module and data outputting module, the output of alimentary canal hysteroscope module and the input of autonomous database connect, autonomous database is bi-directionally connected with image segmentation module, and image segmentation module is connect by data computation module with data outputting module.The invention also discloses the recognition methods using the hysteroscope medical image identifying system based on deep learning.The present invention carries out pattern-recognition to the original image collected by autonomous database and obtains identification image, then it carries out image segmentation by image segmentation module and recognition result image is calculated by data computation module progress regression analysis to be exported, solve the problems, such as that very big deviation occurs in the size for easily leading to calculated original measurement object when manually estimating hysteroscope medical image.
Description
Technical field
The present invention relates to medical domain, specifically a kind of hysteroscope medical image identifying system and its knowledge based on deep learning
Other method.
Background technique
Nowadays, with the rapid development of science and technology, medical domain has also obtained quick development.Wherein, medical image conduct
A technique in order to which medical treatment or medical research to obtain interior tissue image to human body or human body part with non-intruding mode
With treatment process, it is widely applied in medical domain.
Currently, the medical image being often related in gastroscope colonoscopy, and gastroscope colonoscopy checks result
There are a biggish defects for judgement: since lesion size and property need to estimate, causing testing result because of inspection personnel's experience
Difference, imaging is of different sizes and occurs very big deviation when calculating the size of original measurement object, so that patient is not to
It will appear different judgement auxiliary datas with the inspection that medical institutions do, objective and accurate data can not be provided for doctor.Therefore,
The problem of designing a kind of hysteroscope medical image identifying system and its recognition methods based on deep learning, becoming urgent need to resolve.
Summary of the invention
The hysteroscope medical image identifying system and its identification side that the purpose of the present invention is to provide a kind of based on deep learning
It is laggard to carry out image segmentation by autonomous database deep learning method to solve the problems mentioned in the above background technology for method
Enter data and calculate center segmented image area article size is returned to calculate to generate data, variation, which no matter is imaged, much can return
The size of original measurement object is calculated, provides more accurate auxiliary data for subsequent medical behavior.
To achieve the above object, the invention provides the following technical scheme:
A kind of hysteroscope medical image identifying system based on deep learning, comprising: alimentary canal hysteroscope module, autonomous database, figure
As segmentation module, data computation module and data outputting module, the output of the alimentary canal hysteroscope module and autonomous database
Input connection, the autonomous database are bi-directionally connected with image segmentation module, and described image divides module and calculates mould by data
Block is connect with data outputting module；
The alimentary canal hysteroscope module is used to carry out doubtful lesion Image Acquisition to obtain original image and be transmitted to from master data
Library；The autonomous database is used to carry out mode using deep learning method to the collected original image of alimentary canal hysteroscope module
Identification obtains identification image and is transmitted to image segmentation module；Described image divides module and is used to use the identification image of acquisition
The method of deep learning carries out image segmentation and obtains segmented image；The data computation module be used for the segmented image of acquisition into
Row regression analysis is calculated article size data in original image and generates recognition result image；The data outputting module is used
Article size data and recognition result image carry out output printed report in the original image to acquisition；Pass through the original of acquisition
Objects in images size data and recognition result image provide more accurate auxiliary data for subsequent medical behavior, and computer passes through
Self-teaching mode constantly is deepened recognition capability.
As a further solution of the present invention: the alimentary canal hysteroscope module include Fuji can 4450, EG-600WP and
Olympus GIF-Q180 etc..
As further scheme of the invention: the pattern-recognition is to pass through the picture that will be identified in original image
The image initials such as size, shape data compare in the database generates property to judge that image data obtains identification image.
As further scheme of the invention: it includes computer that described image, which divides module,；The requirement of the computer
Are as follows: 32 thread x2 of Inter18 core is saved as in CPU, inside saves as 256G, solid state hard disk 1T, mechanical hard disk 100T, operating system
For LINUX operating system or Windows7 operating system or more.
As further scheme of the invention: described image, which is divided into, passes through computer network progress one for identification image
The convolution of series, nonlinear transformation, down-sampling operate to obtain the image of low resolution, then carry out up-sampling by deconvolution and mention
High image resolution obtains the image segmentation result of time low resolution, obtains finally by the size for being upsampled to former identification image
Segmented image.
As further scheme of the invention: the computer network is broadband network or local area network, passes through computer
Display and computer are attached by network.
As further scheme of the invention: described image divides module by interacting with autonomous database come to segmentation
Image carries out automatic clustering and self-teaching amendment.
As further scheme of the invention: the regression analysis is calculated as big to the image of identification image target area
It is small to carry out calculating the longest diameter for obtaining image and most short diameter and generate area data, then reality is revert to by computation model
Border needs to measure the size of object to obtain article size data in the original image of alimentary canal hysteroscope module acquisition and generate knowledge
Other result images.
As further scheme of the invention: the data outputting module includes display and printer.
A kind of recognition methods of the hysteroscope medical image identifying system using above-mentioned based on deep learning, steps are as follows:
1) doctor obtains original image to doubtful lesion progress Image Acquisition by alimentary canal hysteroscope module and is transmitted to autonomous number
Pattern-recognition, which is carried out, according to library obtains identification image；
2) image segmentation is carried out using the method for deep learning by identification image of the image segmentation module to acquisition to be divided
Then image carries out regression analysis by segmented image of the data computation module to acquisition and the object in original image is calculated
Data simultaneously generate recognition result image；
3) by data outputting module in the original image of acquisition object data and recognition result image carry out output printing
Report.
Application of the hysteroscope medical image identifying system based on deep learning in medical image identification.
Compared with prior art, the beneficial effects of the present invention are:
The present invention is provided with alimentary canal hysteroscope module, autonomous database, image segmentation module, data computation module and data output
Module, by alimentary canal hysteroscope module to doubtful lesion carry out Image Acquisition obtain original image and be transmitted to autonomous database into
Row pattern-recognition obtains identification image, the method for then using deep learning by identification image of the image segmentation module to acquisition
It carries out image segmentation and obtains segmented image, then regression analysis is carried out by segmented image of the data computation module to acquisition and is calculated
Object data into original image simultaneously generates recognition result image and is exported, when solving artificial range estimation hysteroscope medical image
There is the problem of very big deviation in the size for easily leading to calculated original measurement object, and variation, which no matter is imaged, much can return
Return the size for calculating original measurement object, while avoiding because of object to be imaged error in judgement caused by experience is insufficient,
It can be needed with the object model views etc. that swell in extensive utilization to medical treatment, underwater rescue and animal body through image judgment object size
In situation, have a vast market foreground.
Detailed description of the invention
Fig. 1 is the structural block diagram of the hysteroscope medical image identifying system based on deep learning.
Fig. 2 is embodiment 3 based on original image (a), artificial mark in the hysteroscope medical image identifying system of deep learning
Picture (b) and the schematic diagram of AI identification picture (c).
Fig. 3 is schematic diagram of the embodiment 3 based on recognition result image in the hysteroscope medical image identifying system of deep learning.
Specific embodiment
Present invention will be explained in further detail in the following with reference to the drawings and specific embodiments.Following embodiment will be helpful to
Those skilled in the art further understands the present invention, but the invention is not limited in any way.It should be pointed out that ability
For the those of ordinary skill in domain, without departing from the inventive concept of the premise, various modifications and improvements can be made.These
Belong to protection scope of the present invention.
It should be appreciated that ought use in this specification and in the appended claims, term " includes " and "comprising" instruction
Described feature, entirety, step, operation, the presence of element and/or component, but one or more of the other feature, whole is not precluded
Body, step, operation, the presence or addition of element, component and/or its set.
It is also understood that mesh of the term used in this description of the invention merely for the sake of description specific embodiment
And be not intended to limit the present invention.As description of the invention and it is used in the attached claims, unless on
Other situations are hereafter clearly indicated, otherwise " one " of singular, "one" and "the" are intended to include plural form.
Embodiment 1
Referring to Fig. 1, a kind of hysteroscope medical image identifying system based on deep learning, including it is alimentary canal hysteroscope module, autonomous
Database, image segmentation module, data computation module and data outputting module, the hysteroscope medical image based on deep learning
Identifying system further include for system provide stable power-supplying power supply, wherein the output of the alimentary canal hysteroscope module with from
The input of primary database connects, and the autonomous database is bi-directionally connected with image segmentation module, and described image segmentation module passes through
Data computation module is connect with data outputting module, and the alimentary canal hysteroscope module is used to carry out Image Acquisition to doubtful lesion to obtain
To original image and it is transmitted to autonomous database；
In order to which the size for easily leading to calculated original measurement object when solving and manually estimating hysteroscope medical image occurs greatly
The problem of deviation, the alimentary canal hysteroscope module include Fuji's energy 4450, EG-600WP and Olympus GIF-Q180 etc., institute
Autonomous database is stated for carrying out pattern-recognition using deep learning method to the collected original image of alimentary canal hysteroscope module
It obtains identification image and is transmitted to image segmentation module, the pattern-recognition is the picture by will identify in original image
The image initials such as size, shape data compare in the database generates property to judge that image data obtains identification image；
Further, described image segmentation module is used to carry out image point using the method for deep learning to the identification image of acquisition
Cut to obtain segmented image, specifically, described image be divided into will identification image by computer network carry out a series of convolution,
Nonlinear transformation, down-sampling operate to obtain the image of low resolution, then carry out up-sampling by deconvolution and improve image resolution
Rate obtains the image segmentation result of time low resolution, obtains segmented image finally by the size for being upsampled to former identification image,
It includes computer that described image, which divides module, the requirement of the computer are as follows: Inter18 core 32 thread x2, GPU are saved as in CPU
For TITAN XP 12Gx4,256G, solid state hard disk 1T, mechanical hard disk 100T are inside saved as, operating system is LINUX operation system
System or Windows7 operating system or more；
Further, the computer network is broadband network, is attached display and computer by computer network,
Described image divides module and carries out automatic clustering and self-teaching amendment to segmented image by interacting with autonomous database；
Further, the data computation module is used to carry out regression analysis to the segmented image of acquisition that original image to be calculated
Middle article size data simultaneously generate recognition result image, and the regression analysis is calculated as big to the image of identification image target area
It is small to carry out calculating the longest diameter for obtaining image and most short diameter and generate area data, then reality is revert to by computation model
Border needs to measure the size of object to obtain article size data in the original image of alimentary canal hysteroscope module acquisition and generate knowledge
Other result images；
Further, the data outputting module is used for article size data in the original image of acquisition and recognition result image
Output printed report is carried out, the original graph of acquisition can also be passed through by printing images onto A4 paper (measurement of default full width)
Article size data and recognition result image are that subsequent medical behavior provides more accurate auxiliary data as in, and computer passes through oneself
My mode of learning constantly deepens recognition capability；
Further, the data outputting module includes display and printer, and the autonomous database includes memory, described
Memory can be realized by any kind of volatibility or non-volatile memory device or their combination, as static random is deposited
Access to memory, electrically erasable programmable read-only memory, Erasable Programmable Read Only Memory EPROM, programmable read only memory, only
Read memory, magnetic memory, flash memory, disk or CD.
In the present embodiment, a kind of identification side of the hysteroscope medical image identifying system using above-mentioned based on deep learning
Method, steps are as follows:
1) doctor obtains original image to doubtful lesion progress Image Acquisition by alimentary canal hysteroscope module and is transmitted to autonomous number
Pattern-recognition, which is carried out, according to library obtains identification image；
2) image segmentation is carried out using the method for deep learning by identification image of the image segmentation module to acquisition to be divided
Then image carries out regression analysis by segmented image of the data computation module to acquisition and the object in original image is calculated
Data simultaneously generate recognition result image；
3) by data outputting module in the original image of acquisition object data and recognition result image carry out output printing
Report.
In the present embodiment, a kind of computer readable storage medium is stored thereon with computer program instructions, the program instruction
The step of above method is realized when being executed by processor.
In the present embodiment, the hysteroscope medical image identifying system based on deep learning is in medical image identification
Using.
Embodiment 2
Referring to Fig. 1, a kind of hysteroscope medical image identifying system based on deep learning, comprising: alimentary canal hysteroscope module, autonomous
Database, image segmentation module, data computation module and data outputting module, the output of the alimentary canal hysteroscope module and autonomous
The input of database connects, and the autonomous database is bi-directionally connected with image segmentation module, and described image segmentation module passes through number
It is connect according to computing module with data outputting module；
In order to which the size for easily leading to calculated original measurement object when solving and manually estimating hysteroscope medical image occurs greatly
The problem of deviation, the alimentary canal hysteroscope module are used to carry out doubtful lesion Image Acquisition to obtain original image and be transmitted to certainly
Primary database；The autonomous database be used for the collected original image of alimentary canal hysteroscope module using deep learning method into
Row pattern-recognition obtains identification image and is transmitted to image segmentation module；Described image divides module for the identification figure to acquisition
Segmented image is obtained as the method using deep learning carries out image segmentation；The data computation module is for the segmentation to acquisition
Image carries out regression analysis and article size data in original image is calculated and generate recognition result image；The data output
Module can also lead to for carrying out output printed report to article size data in the original image of acquisition and recognition result image
It crosses and prints images onto A4 paper (measurement of default full width)；Pass through article size data and recognition result in the original image of acquisition
Image provides more accurate auxiliary data for subsequent medical behavior, and computer constantly is deepened to identify by self-teaching mode
Ability；
Further, the pattern-recognition is by by image initials such as the picture size identified in original image, shapes
Data compare in the database generates property to judge that image data obtains identification image, and described image, which is divided into, will identify image
A series of convolution, nonlinear transformation, down-sampling are carried out by computer network to operate to obtain the image of low resolution, are then led to
It crosses deconvolution to carry out up-sampling and improve image resolution ratio obtaining the image segmentation result of time low resolution, finally by being upsampled to
The size of original identification image obtains segmented image, the regression analysis be calculated as to the image size of identification image target area into
Row calculates the longest diameter for obtaining image and most short diameter and generates area data, then revert to practical need by computation model
The size of object is measured to obtain article size data in the original image of alimentary canal hysteroscope module acquisition and generate identification knot
Fruit image.
Preferably, described image segmentation module includes computer；The requirement of the computer are as follows: Inter18 is saved as in CPU
Core 32 thread x2, GPU are TITAN XP 12Gx4, inside save as 256G, solid state hard disk 1T, mechanical hard disk 100T, operation system
System is LINUX operating system or Windows7 operating system or more.
Preferably, the computer network is local area network, is attached display and computer by computer network,
Described image divides module and carries out automatic clustering and self-teaching amendment to segmented image by interacting with autonomous database.
A kind of recognition methods of the hysteroscope medical image identifying system using above-mentioned based on deep learning, steps are as follows:
1) doctor obtains original image to doubtful lesion progress Image Acquisition by alimentary canal hysteroscope module and is transmitted to autonomous number
Pattern-recognition, which is carried out, according to library obtains identification image；
2) image segmentation is carried out using the method for deep learning by identification image of the image segmentation module to acquisition to be divided
Then image carries out regression analysis by segmented image of the data computation module to acquisition and the object in original image is calculated
Data simultaneously generate recognition result image；
3) by data outputting module in the original image of acquisition object data and recognition result image carry out output printing
Report.
In the present embodiment, a kind of computer readable storage medium is stored thereon with computer program instructions, the program instruction
The step of above method is realized when being executed by processor.
In the present embodiment, the hysteroscope medical image identifying system based on deep learning is in medical image identification
Using.
Embodiment 3
Please refer to Fig. 1-3, a kind of hysteroscope medical image identifying system based on deep learning, comprising: alimentary canal hysteroscope module, from
Primary database, image segmentation module, data computation module and data outputting module, wherein the alimentary canal hysteroscope module it is defeated
It is connect out with the input of autonomous database, the autonomous database is bi-directionally connected with image segmentation module, and described image divides mould
Block is connect by data computation module with data outputting module, specific as shown in Figure 1；
Further, the alimentary canal hysteroscope module is used to carry out doubtful lesion Image Acquisition to obtain original image and be transmitted to
Autonomous database；The alimentary canal hysteroscope module includes Fuji's energy 4450, EG-600WP and Olympus GIF-Q180 etc.；Institute
Autonomous database is stated for carrying out pattern-recognition using deep learning method to the collected original image of alimentary canal hysteroscope module
It obtains identification image and is transmitted to image segmentation module；The pattern-recognition is the picture by will identify in original image
The image initials such as size, shape data compare in the database generates property to judge that image data obtains identification image；
Further, described image segmentation module is used to carry out image point using the method for deep learning to the identification image of acquisition
It cuts to obtain segmented image；It includes computer that described image, which divides module,；The requirement of the computer are as follows: Inter18 is saved as in CPU
Core 32 thread x2, GPU are TITAN XP 12Gx4, inside save as 256G, solid state hard disk 1T, mechanical hard disk 100T, operation system
System is LINUX operating system or Windows7 operating system or more；Described image, which is divided into, passes through computer for identification image
Network carries out a series of convolution, nonlinear transformation, down-sampling and operates to obtain the image of low resolution, then by deconvolution into
Up-sampling of going improves image resolution ratio and obtains the image segmentation result of time low resolution, identifies image finally by original is upsampled to
Size obtain segmented image；
Further, the computer network is local area network, is attached display and computer by computer network, institute
It states image segmentation module and automatic clustering and self-teaching amendment is carried out to segmented image by interacting with autonomous database；It is described
Data computation module is used to carry out the segmented image of acquisition regression analysis and article size data is calculated in original image simultaneously
Recognition result image is generated, the regression analysis is calculated as carrying out calculating to the image size of identification image target area obtaining figure
The longest diameter of picture and most short diameter simultaneously generate area data, then revert to actual needs measurement object by computation model
Size obtains article size data and generating recognition result image in the original image of alimentary canal hysteroscope module acquisition；
Further, the data outputting module is used for article size data in the original image of acquisition and recognition result image
Output printed report is carried out, it can also be by printing images onto A4 paper (measurement of default full width)；Pass through the original graph of acquisition
Article size data and recognition result image are that subsequent medical behavior provides more accurate auxiliary data as in, and computer passes through oneself
My mode of learning constantly deepens recognition capability.
In the present embodiment, a kind of identification side of the hysteroscope medical image identifying system using above-mentioned based on deep learning
Method, steps are as follows:
1) doctor obtains original image (original image) to doubtful lesion progress Image Acquisition by alimentary canal hysteroscope module and passes
It transports to autonomous database progress pattern-recognition and obtains identification image (AI identifies picture), identify image for breath by pattern-recognition
Meat, specifically as shown in Figure 2, wherein (a) is original image, (b) for artificial mark picture (i.e. using traditional manual method into
Row identification mark), (c) picture is identified for AI；
2) identify that picture carries out image segmentation using the method for deep learning and divided by AI of the image segmentation module to acquisition
Image is cut, regression analysis is then carried out by segmented image of the data computation module to acquisition, the object in original image is calculated
Volume data simultaneously generates recognition result image；Specifically as shown in Figure 3, by calculate actual object longest diameter corresponding to image and
Most short diameter, and reference area is generated, real data is calculated according to regression formula, determines that coefficient is 0.93277264, it is comprehensive
Above-mentioned characteristic is 98.7% by the accuracy rate that 10,000 pictures carry out calculating identification；
3) by data outputting module in the original image of acquisition object data and recognition result image carry out output printing
Report can print images onto A4 paper (measurement of default full width).
In the present embodiment, by returning object judgement, while applying to intracavitary and epidermis is exposed with intracavitary striograph and CT
Outer occupy-place block calculates, and is 99.9% with CT picture coincidence rate.
In the present embodiment, the hysteroscope medical image identifying system based on deep learning is in medical image identification
Using.
The working principle of the invention is: carrying out Image Acquisition to doubtful lesion by alimentary canal hysteroscope module and obtains original graph
As and be transmitted to autonomous database carry out pattern-recognition obtain identification image, by image segmentation module to the identification image of acquisition
Image segmentation is carried out using the method for deep learning and obtains segmented image, then by data computation module to the segmentation figure of acquisition
It is exported, no matter is imaged as progress regression analysis is calculated the object data in original image and generates recognition result image
Change it is much can calculated original measurement object size, while avoid because caused by experience is insufficient by
As object error in judgement, can need to pass through figure with the object model views etc. that swell in extensive utilization to medical treatment, underwater rescue and animal body
Piece judgment object exist with by the case where image judgment object size.
The beneficial effects of the present invention are: the present invention is provided with alimentary canal hysteroscope module, autonomous database, image segmentation mould
Block, data computation module and data outputting module, doctor carry out Image Acquisition to doubtful lesion by alimentary canal hysteroscope module and obtain
To original image and be transmitted to autonomous database carry out pattern-recognition obtain identification image, by image segmentation module to acquisition
Identification image carries out image segmentation using the method for deep learning and obtains segmented image, then by data computation module to acquisition
Segmented image carry out regression analysis and the object data in original image is calculated and generates recognition result image is exported,
There is very big deviation in the size for easily leading to calculated original measurement object when solving artificial range estimation hysteroscope medical image
Problem has a vast market foreground.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a computer-readable storage medium
In, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the storage medium can for
Machine memory, flash memory, read-only memory, programmable read only memory, electrically erasable programmable memory, register etc..
Better embodiment of the invention is explained in detail above, but the present invention is not limited to above-mentioned embodiment party
Formula within the knowledge of one of ordinary skill in the art can also be without departing from the purpose of the present invention
Various changes can be made.There is no necessity and possibility to exhaust all the enbodiments.And it thus amplifies out apparent
Variation or variation be still in the protection scope of this invention.
Claims (10)
1. a kind of hysteroscope medical image identifying system based on deep learning, including alimentary canal hysteroscope module, which is characterized in that also
It include: autonomous database, image segmentation module, data computation module and data outputting module；Wherein
The output of the alimentary canal hysteroscope module and the input of autonomous database connect, the autonomous database and image segmentation mould
Block is bi-directionally connected, and image segmentation module is connect by data computation module with data outputting module；
The alimentary canal hysteroscope module is used to carry out doubtful lesion Image Acquisition to obtain original image and be transmitted to from master data
Library；
The autonomous database is used to carry out mould using deep learning method to the collected original image of alimentary canal hysteroscope module
Formula identifies to obtain identification image and is transmitted to image segmentation module；
Described image segmentation module is used to divide the identification image of acquisition using the method progress image segmentation of deep learning
Cut image；
The data computation module is used to that object in original image to be calculated to the segmented image progress regression analysis of acquisition big
Small data simultaneously generates recognition result image；
The data outputting module is for exporting article size data in the original image of acquisition and recognition result image.
2. the hysteroscope medical image identifying system according to claim 1 based on deep learning, which is characterized in that described to disappear
Changing road hysteroscope module includes Fuji's energy 4450, EG-600WP and Olympus GIF-Q180 etc..
3. the hysteroscope medical image identifying system according to claim 1 or 2 based on deep learning, which is characterized in that institute
Stating pattern-recognition is by including the image initial data of picture size and shape in data by what is identified in original image
It is compared in library and generates property to judge image data and obtain identification image.
4. the hysteroscope medical image identifying system according to claim 3 based on deep learning, which is characterized in that the figure
As segmentation module includes computer.
5. the hysteroscope medical image identifying system according to claim 4 based on deep learning, which is characterized in that the figure
Identification image is operated to obtain low resolution by computer network progress convolution, nonlinear transformation and down-sampling as being divided into
Then image carries out up-sampling raising image resolution ratio by deconvolution and obtains time low-resolution image, finally by up-sampling
Size to former identification image obtains segmented image.
6. the hysteroscope medical image identifying system according to claim 5 based on deep learning, which is characterized in that the meter
Calculation machine network is broadband network or local area network.
7. the hysteroscope medical image identifying system according to claim 6 based on deep learning, which is characterized in that described time
Returning analytical calculation is to carry out the longest diameter and most short diameter that measuring and calculating obtains image to the image size of identification image target area
And area data is generated, it revert to the size of actual needs measurement object by computation model then to obtain object in original image
Body size data simultaneously generates recognition result image.
8. the hysteroscope medical image identifying system according to claim 7 based on deep learning, which is characterized in that the number
It include display and printer according to output module.
9. a kind of identification of the hysteroscope medical image identifying system using a method as claimed in any one of claims 1-8 based on deep learning
Method, which is characterized in that steps are as follows:
1) doctor obtains original image to doubtful lesion progress Image Acquisition by alimentary canal hysteroscope module and is transmitted to autonomous number
Pattern-recognition, which is carried out, according to library obtains identification image；
2) image segmentation is carried out using the method for deep learning by identification image of the image segmentation module to acquisition to be divided
Then image carries out regression analysis by segmented image of the data computation module to acquisition and the object in original image is calculated
Data simultaneously generate recognition result image；
3) by data outputting module in the original image of acquisition object data and recognition result image carry out output printing
Report.
10. a kind of hysteroscope medical image identifying system a method as claimed in any one of claims 1-8 based on deep learning is in medicine shadow
As the application in identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811599884.1A CN109754081A (en) | 2018-12-26 | 2018-12-26 | A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811599884.1A CN109754081A (en) | 2018-12-26 | 2018-12-26 | A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109754081A true CN109754081A (en) | 2019-05-14 |
Family
ID=66403903
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811599884.1A Withdrawn CN109754081A (en) | 2018-12-26 | 2018-12-26 | A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109754081A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062945A (en) * | 2019-03-28 | 2020-04-24 | 芨影(厦门)科技有限公司 | Early cancer medical image and genotype AI recognition system and method and application thereof |
| CN112241713A (en) * | 2020-10-22 | 2021-01-19 | 江苏美克医学技术有限公司 | Vaginal microorganism identification method and device based on pattern identification and deep learning |
-
2018
- 2018-12-26 CN CN201811599884.1A patent/CN109754081A/en not_active Withdrawn
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062945A (en) * | 2019-03-28 | 2020-04-24 | 芨影(厦门)科技有限公司 | Early cancer medical image and genotype AI recognition system and method and application thereof |
| CN112241713A (en) * | 2020-10-22 | 2021-01-19 | 江苏美克医学技术有限公司 | Vaginal microorganism identification method and device based on pattern identification and deep learning |
| CN112241713B (en) * | 2020-10-22 | 2023-12-29 | 江苏美克医学技术有限公司 | Method and device for identifying vaginal microorganisms based on pattern recognition and deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Claes et al. | Computerized craniofacial reconstruction: conceptual framework and review | |
| Looney et al. | Automatic 3D ultrasound segmentation of the first trimester placenta using deep learning | |
| CN108022238A (en) | Method, computer-readable storage medium and the system being detected to object in 3D rendering | |
| Kohlmann et al. | Automatic lung segmentation method for MRI-based lung perfusion studies of patients with chronic obstructive pulmonary disease | |
| Tilotta et al. | Construction and analysis of a head CT-scan database for craniofacial reconstruction | |
| WO2012174495A2 (en) | Physics based image processing and evaluation process of perfusion images from radiology imaging | |
| JP6888041B2 (en) | How to get a medical sagittal image, how to train a neural network to get a medical sagittal image, and a computer device | |
| CN111415361B (en) | Method and device for estimating brain age of fetus and detecting abnormality based on deep learning | |
| Xu et al. | Fetal pose estimation in volumetric MRI using a 3D convolution neural network | |
| CN102622759A (en) | Gray scale and geometric information combined medical image registration method | |
| JP2015033556A (en) | Image processing system and image processing method | |
| CN110163877A (en) | A kind of method and system of MRI ventricular structure segmentation | |
| CN111311703A (en) | Electrical impedance tomography image reconstruction method based on deep learning | |
| CN104545999B (en) | Method and device for measuring bladder volume through ultrasound images | |
| CN109754081A (en) | A kind of hysteroscope medical image identifying system and its recognition methods based on deep learning | |
| CN114565572A (en) | Cerebral hemorrhage CT image classification method based on image sequence analysis | |
| KR101790437B1 (en) | Method and apparatus for processing medical image | |
| Liu et al. | Automated classification and measurement of fetal ultrasound images with attention feature pyramid network | |
| Hameeteman et al. | Carotid wall volume quantification from magnetic resonance images using deformable model fitting and learning-based correction of systematic errors | |
| Zeng et al. | TUSPM-NET: A multi-task model for thyroid ultrasound standard plane recognition and detection of key anatomical structures of the thyroid | |
| Haggerty et al. | Quantification of left ventricular volumes, mass, and ejection fraction using cine displacement encoding with stimulated echoes (DENSE) MRI | |
| CN111062945A (en) | Early cancer medical image and genotype AI recognition system and method and application thereof | |
| Yan et al. | A quantitative description of pelvic floor muscle fibre organisation | |
| CN114708283A (en) | Image object segmentation method and device, electronic equipment and storage medium | |
| WO2014106747A1 (en) | Methods and apparatus for image processing |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190514 |
|
| WW01 | Invention patent application withdrawn after publication |