CN110245657A - Pathological image similarity detection method and detection device - Google Patents
Pathological image similarity detection method and detection device Download PDFInfo
- Publication number
- CN110245657A CN110245657A CN201910411362.2A CN201910411362A CN110245657A CN 110245657 A CN110245657 A CN 110245657A CN 201910411362 A CN201910411362 A CN 201910411362A CN 110245657 A CN110245657 A CN 110245657A
- Authority
- CN
- China
- Prior art keywords
- image
- pathological image
- pathological
- interest
- convolutional neural
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06V10/267—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 by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Primary Health Care (AREA)
- Biophysics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Epidemiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Databases & Information Systems (AREA)
Abstract
A kind of pathological image similarity detection method and detection device, detection method include: acquisition pathological image, are labeled at least one area-of-interest on pathological image, and by the area-of-interest classified finishing of each mark；The area-of-interest of all marks is divided into two parts, a part is used as training image, and another part is used to establish pathological image library；It is multiple segments by the prospect cutting of training image, distributes classification same as affiliated area-of-interest for segment；There is the feature in the segment of class label and classification using depth convolutional neural networks model extraction, thus training depth convolutional neural networks model；Test image is input in trained depth convolutional neural networks model, depth convolutional neural networks model exports classification belonging to the test image, and according to the test image generic, the pathological image of respective classes is searched in pathological image library.This method can quickly improve the diagosis ability of pathologist and the confidence level of diagosis result.
Description
Technical field
The present invention relates to artificial intelligence fields, specifically, are related to pathological image similarity detection method and detection device.
Background technique
In clinical medicine, pathology are one of the most important methods to diagnose the illness.Pathologic finding also tends to be referred to as
The goldstandard of medical diagnosis on disease.Clinician is generally basede on experience and knowledge, makes empirical diagnosis, and final diagnosis is often
Pathologist Binding experiment is needed to make.It generally requires to take out one piece of tissue from patient's affected area, preparation slice, and in electronics
The form of microscopically observation cell, tissue finally provides pathological replacement and provides diagnostic comments.The incubation of pathologist
It is very long and uninteresting.The pathologist qualified as one generally requires the several years, strict training in even 10 years could be completed.One
Position pathologist, preliminary pathological replacement can just be provided by needing conscientiously to peruse 10,000 or more pathological sections.
With the development of high-resolution medical image imaging technique, digitizes pathology system and obtained in worldwide extensively
General application.Digitlization pathology system improves the diagosis efficiency of pathologist, while also facilitating the preservation and filing of data.
And in terms of pathologist culture, typical pathological section is carried out to build library using digitlization pathology system, can greatly be mentioned
The culture efficiency for rising pathologist, shortens the cultivation cycle of pathologist.Using digitlization pathological images library, pathologist can be with
Unknown slice is compared with slice in library easily and fast, consults typical case, this is conducive to quickly improve pathologist
Diagosis ability and diagosis result confidence level.Therefore, the similar area detection of pathological image becomes digitlization pathology shadow
As an important problem in analysis field.And computer based similar area detection method, subjective factors can be cut down
Influence, it is quick, accurate and reproducible to have the characteristics that.
Summary of the invention
The present invention is automatically analyzed digitlization pathological images using convolutional neural networks and feature extraction, passes through inspection
The similarity with known segment is surveyed, corresponding classification information is provided, and this classification information can be used in retrieving digitized pathology library
Segment in corresponding classification, it is quick, accurate and reproducible to have the characteristics that.Concrete scheme is as follows:
A kind of pathological image similarity detection method, comprising:
Step S1 obtains multiple pathological images, is labeled extraction at least one area-of-interest on pathological image, and
By the area-of-interest classified finishing of each mark；
The area-of-interest of all marks is divided into two parts by step S2 in proportion, and a part is used as training image, separately
It is a part of then be used to establish pathological image library；
The prospect of training image is split by step S3 with background, and is multiple segments by the prospect cutting being partitioned into,
Classification same as affiliated area-of-interest is distributed for segment, to being cut into the pathological image of multiple with class label
Multiple segments；
Step S4, using the feature in depth convolutional neural networks model extraction segment and classification, thus the training depth
Spend convolutional neural networks model；
Test image is input in trained depth convolutional neural networks model by step S5, depth convolutional Neural net
Network model exports classification belonging to the test image, also, the classification according to belonging to the test image, establishes in step s 2
Pathological image and the display of respective classes are searched in pathological image library.
Preferably, the depth convolutional neural networks model is inception-resnet-v2 network model, including successively
The Input layer of connection, Stem layers, Inception-A layers of 5x, Reduction-A layers, Inception-B layers of 10x,
Reduction-B layers, Inception-C layers of 5x, Pooling layers, Dropout layers, Softmax layers of Average.
Preferably, in step s3, multiple segments with class label are overturn using Random Level or vertically, is random
Rotation, random cropping increase the diversity of data set.
Preferably, the optimizer of depth convolutional neural networks is RMSProp.
Preferably, the hyper parameter of depth convolutional neural networks model includes learning rate, learning strategy, weight attenuation coefficient,
Optimal model hyper parameter is chosen using grid search.
Preferably, the boundary that area-of-interest is marked using polygon annotation tool, by the boundary of the area-of-interest
Corresponding classification is collectively stored in the file of xml format.
Preferably, the ratio in step S2 is not less than 9:1.
Preferably, the segment is having a size of no more than 320x320 pixel.
Preferably, the feature extracted in step S4 includes low-level features and advanced features, the low-level features include point,
Line, edge, the advanced features include tissue morphology.
The present invention also provides a kind of pathological image similitude detection devices, comprising:
Mark categorization module carries out at least one area-of-interest on pathological image for obtaining multiple pathological images
Mark, and by the area-of-interest classified finishing of each mark；
Division module, for the area-of-interest of all marks to be divided into two parts in proportion, a part is as training
Image, another part are then used to establish pathological image library；
Foreground segmentation module, the prospect cutting for the prospect of training image to be split with background, and will be partitioned into
For multiple segments；
Label determining module assigns classification same as affiliated area-of-interest for segment, thus by the pathology figure of multiple
As being cut into multiple segments with class label；
Training module, using the feature in depth convolutional neural networks model extraction segment and classification, thus described in training
Depth convolutional neural networks model；
Application module, for test image to be input in trained depth convolutional neural networks model, depth convolution
Neural network model exports classification belonging to the test image, also, the classification according to belonging to the test image, in pathological image
Pathological image and the display of respective classes are searched in library.
This method can fast and accurately analyze digitlization pathological images, can be realized and digitlization pathology shadow
As image compares and detects similar area in library, to quickly improve pathologist diagosis ability and diagosis result can
Reliability, the present processes are highly suitable for the analysis of organ pathological images, are particularly suitable for the analysis of lung pathologies image.
Detailed description of the invention
By the way that embodiment is described in conjunction with following accompanying drawings, features described above of the invention and technological merit will become
More understands and be readily appreciated that.
Fig. 1 is the step schematic diagram for indicating the pathological image similarity detection method of the embodiment of the present invention；
Fig. 2 is the schematic network structure for indicating the inception resnet v2 of the embodiment of the present invention；
Fig. 3 is the process signal for indicating the pathological image similarity detection method of the embodiment of the present invention；
Fig. 4 is the hardware structure schematic diagram for indicating the pathological image similitude detection device of the embodiment of the present invention.
Specific embodiment
The reality of pathological image similarity detection method of the present invention and detection device described below with reference to the accompanying drawings
Apply example.Those skilled in the art will recognize, without departing from the spirit and scope of the present invention, Ke Yiyong
A variety of different modes or combinations thereof are modified described embodiment.Therefore, attached drawing and description are inherently explanation
Property, it is not intended to limit the scope of the claims.In addition, in the present specification, attached drawing is drawn not in scale, and
Identical appended drawing reference indicates identical part.
Fig. 1 shows the step schematic diagram of pathological image similarity detection method in the present embodiment, and Fig. 3 is the embodiment of the present invention
Pathological image similarity detection method flow diagram.Illustrate that pathological image similitude is detected below with reference to Fig. 1, Fig. 3
Method comprising following steps:
Step S1 obtains multiple pathological images, carries out data mark to these pathological images, can be by experienced disease
Manage doctor or expert at least one interested area-of-interest during pathological image upper ledge selects doctor in practical diagosis
(ROI), annotation callout is carried out on region.Also, also further by each tab area classified finishing.Polygon mark can be used
Note tool marks the boundary of ROI, its classification and boundary are collectively stored in the file of an xml format.For example, some stomach
In portion's pathological image, there is polyp, tumour, the different situations of erosion, then it can frame selects polyp respectively by the stomach pathological image
Region, tumour region, rotten to the corn region, and marked respectively.Then, classify for the ROI of all pathological images,
It is hereby achieved that multiple classification, such as stomach polyp regions class, esophageal cyst region class etc..
The data marked are divided into two parts by step S2 in proportion, it is preferable that using the ratio for being not less than 9:1.Compared with
Training of more part data as training image, for subsequent similar area detection algorithm model；Less part is then
For establishing pathological image library, high-quality, representational data are chosen preferentially to establish pathological image library, compared for doctor,
With reference to.
Step S3, training image may include the blank background of large area, therefore prospect and background segment can be come out.
For the foreground part being partitioned into, may still have biggish size, can be again multiple segments by prospect cutting.It is preferred that
Ground, the segment after setting cutting is having a size of no more than 320x320 pixel.
Classification same as affiliated area-of-interest is distributed for segment, so as to cut whole digitlization pathological images
It is divided into numerous segments, and every segment has its corresponding class label.So far available that largely there is classification mark
The segment of label, for being trained to subsequent neural network model.
Step S4 trains neural network model using largely having the segment of class label obtained in step S3,
The neural network model extracts low-level features and advanced features in segment, low-level features point, line, edge etc., advanced features
Including tissue morphology etc., and these features to be used for final classification.By the feature extraction and classification of a large amount of segment come not
The disconnected nicety of grading for improving neural network model.
Test image is input in trained neural network model by step S5, by calculating, neural network mould
Type can export classification belonging to the test image.Can be according to the classification of the output, the number that has built up in step s 2
Change the pathological images that respective classes are searched in pathological images library, for doctor's comparison, reference.
In one alternate embodiment, in step S4, the neural network model is depth convolutional neural networks model, more
It specifically, is inception-resnet-v2 network model, corresponding network structure is as shown in Fig. 2, include sequentially connected
Input layers, Stem layers, Inception-A layers of 5x, Reduction-A layers, Inception-B layers of 10x, Reduction-B
Layer, Inception-C layers of 5x, Average Pooling (average pond) layer, Dropout layers, Softmax layers.inception
Resnet v2 merges (resnet) residual error network with the network structure of inception, can further be lifted at image classification
On accuracy rate.
It in one alternate embodiment,, can be by multiple tools to increase amount of training data in the model training of step S4
There is the segment of class label to carry out data amplification using image pre-processing method.For example, using Random Level or vertical overturning, with
Machine rotation, random cropping increase the diversity of data set, such as the segment of 320x320 pixel are cut to 299x299 at random
The segment of pixel.
In one alternate embodiment, the optimizer of depth convolutional neural networks is RMSProp (root mean square backpropagation).
In one alternate embodiment, the hyper parameter of depth convolutional neural networks model includes learning rate, learning strategy, power
Weight attenuation coefficient.Optimal model hyper parameter is chosen using grid search.Specifically, it is for different hyper parameter combinations,
Corresponding model is respectively trained, by test model performance, selects optimal hyper parameter combination.
The present invention also provides a kind of pathological image similitude detection device 50, the detection device 50 is that one kind can be according to
The instruction for being previously set or storing, the automatic equipment for carrying out numerical value calculating and/or information processing.For example, it may be intelligent hand
Machine, tablet computer, laptop, desktop computer, rack-mount server, blade server, tower server or cabinet
Formula server (including server cluster composed by independent server or multiple servers) etc..As shown in figure 4, described
Detection device 50 includes following module, and each module can be computer program stored in memory, and memory passes through processing
Device runs the computer program.The memory can be internal storage, such as hard disk or memory.It is also possible to external storage
Device, such as plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital,
SD) block, flash card (Flash Card) etc..
The processor can be central processing unit (Central Processing Unit, CPU), controller, microcontroller
Device, microprocessor or other data processing chips.The processor is used to control the overall operation of the detection device 50, such as
Execute control relevant to the detection device 50 progress data interaction or communication and processing etc..In the present embodiment, the place
Reason device is for running the program code stored in the memory or processing data, such as to run the pathological image similar
Property detection program.
Each module that detection device 50 includes is as follows:
Mark categorization module 501, for obtaining multiple pathological images, at least one area-of-interest on pathological image into
Rower note, and by the area-of-interest classified finishing of each mark；
Division module 502, for the area-of-interest of all marks to be divided into two parts in proportion, a part is as instruction
Practice image, another part is then used to establish pathological image library；
Foreground segmentation module 503 for the prospect of training image to be split with background, and the prospect being partitioned into is cut
It is divided into multiple segments；
Label determining module 504 assigns classification same as affiliated area-of-interest for segment, thus by the pathology of multiple
Image is cut into multiple segments with class label；
Training module 505, using the feature in depth convolutional neural networks model extraction segment and classification, thus training institute
State depth convolutional neural networks model；
Application module 506, for test image to be input in trained depth convolutional neural networks model, depth volume
Product neural network model exports classification belonging to the test image, also, the classification according to belonging to the test image, in pathology figure
Pathological image and display as searching for respective classes in library.
The above description is only a preferred embodiment of the present invention, is not intended to restrict the invention, for those skilled in the art
For member, the invention may be variously modified and varied.All within the spirits and principles of the present invention, it is made it is any modification,
Equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of pathological image similarity detection method characterized by comprising
Step S1 obtains multiple pathological images, is labeled extraction at least one area-of-interest on pathological image, and will be each
The area-of-interest classified finishing of mark；
The area-of-interest of all marks is divided into two parts by step S2 in proportion, and a part is used as training image, another portion
Divide and is then used to establish pathological image library；
The prospect of training image is split by step S3 with background, and is multiple segments by the prospect cutting being partitioned into, for figure
Block distributes classification same as affiliated area-of-interest, to being cut into the pathological image of multiple with the multiple of class label
Segment；
Step S4, using the feature in depth convolutional neural networks model extraction segment and classification, thus the training depth volume
Product neural network model；
Test image is input in trained depth convolutional neural networks model by step S5, depth convolutional neural networks mould
Type exports classification belonging to the test image, also, the classification according to belonging to the test image, the pathology established in step s 2
Pathological image and the display of respective classes are searched in image library.
2. pathological image similarity detection method according to claim 1, which is characterized in that
The depth convolutional neural networks model is inception-resnet-v2 network model, including sequentially connected Input
Layer, Stem layers, Inception-A layers of 5x, Reduction-A layers, Inception-B layers of 10x, Reduction-B layers, 5x
Inception-C layers, Pooling layers, Dropout layers, Softmax layers of Average.
3. pathological image similarity detection method according to claim 1, which is characterized in that
In step s3, multiple segments with class label are used into Random Level or vertical overturning, Random-Rotation, random sanction
It cuts to increase the diversity of data set.
4. pathological image similarity detection method according to claim 1, which is characterized in that
The optimizer of depth convolutional neural networks is RMSProp.
5. pathological image similarity detection method according to claim 1, which is characterized in that
The hyper parameter of depth convolutional neural networks model includes learning rate, learning strategy, weight attenuation coefficient, uses grid search
To choose optimal model hyper parameter.
6. pathological image similarity detection method according to claim 1, which is characterized in that use polygon annotation tool
The corresponding classification in the boundary of the area-of-interest is collectively stored in xml format by the boundary for marking area-of-interest
In file.
7. pathological image similarity detection method according to claim 1, which is characterized in that the ratio in step S2 is not
Less than 9:1.
8. pathological image similarity detection method according to claim 1, which is characterized in that the segment is having a size of little
In 320x320 pixel.
9. pathological image similarity detection method according to claim 1, which is characterized in that
The feature extracted in step S4 includes low-level features and advanced features, and the low-level features include point, line, edge, described
Advanced features include tissue morphology.
10. a kind of pathological image similitude detection device characterized by comprising
Mark categorization module is labeled at least one area-of-interest on pathological image for obtaining multiple pathological images,
And by the area-of-interest classified finishing of each mark；
Division module, for the area-of-interest of all marks to be divided into two parts in proportion, a part is used as training image,
Another part is then used to establish pathological image library；
Foreground segmentation module is more for the prospect of training image to be split with background, and by the prospect cutting being partitioned into
A segment；
Label determining module assigns classification same as affiliated area-of-interest for segment, so that the pathological image of multiple be cut
It is divided into multiple segments with class label；
Training module, using the feature in depth convolutional neural networks model extraction segment and classification, thus the training depth
Convolutional neural networks model；
Application module, for test image to be input in trained depth convolutional neural networks model, depth convolutional Neural
Network model exports classification belonging to the test image, also, the classification according to belonging to the test image, in pathological image library
Search for the pathological image of respective classes and display.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910411362.2A CN110245657B (en) | 2019-05-17 | 2019-05-17 | Pathological image similarity detection method and detection device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910411362.2A CN110245657B (en) | 2019-05-17 | 2019-05-17 | Pathological image similarity detection method and detection device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110245657A true CN110245657A (en) | 2019-09-17 |
| CN110245657B CN110245657B (en) | 2021-08-24 |
Family
ID=67884487
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910411362.2A Active CN110245657B (en) | 2019-05-17 | 2019-05-17 | Pathological image similarity detection method and detection device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110245657B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110866893A (en) * | 2019-09-30 | 2020-03-06 | 中国科学院计算技术研究所 | Pathological image-based TMB classification method and system and TMB analysis device |
| CN111091539A (en) * | 2019-12-09 | 2020-05-01 | 上海联影智能医疗科技有限公司 | Network model training method, medical image processing method, device, medium and equipment |
| CN111368923A (en) * | 2020-03-05 | 2020-07-03 | 上海商汤智能科技有限公司 | Neural network training method and device, electronic equipment and storage medium |
| CN111553361A (en) * | 2020-03-19 | 2020-08-18 | 四川大学华西医院 | Pathological section label identification method |
| CN112287843A (en) * | 2020-10-30 | 2021-01-29 | 西安工程大学 | Doctor and nurse interaction gesture recognition method in operation process |
| CN112614119A (en) * | 2020-12-28 | 2021-04-06 | 上海市精神卫生中心（上海市心理咨询培训中心） | Medical image region-of-interest visualization method, device, storage medium and equipment |
| CN114419314A (en) * | 2020-10-09 | 2022-04-29 | 通用电气精准医疗有限责任公司 | System and method for automatically labeling medical images |
| TWI779284B (en) * | 2020-05-06 | 2022-10-01 | 商之器科技股份有限公司 | Device for marking image data |
| CN117115183A (en) * | 2023-09-01 | 2023-11-24 | 北京透彻未来科技有限公司 | Interception method and system based on digital pathological image visual area |
| WO2024016691A1 (en) * | 2022-07-22 | 2024-01-25 | 华为云计算技术有限公司 | Image retrieval method and apparatus, model training method and apparatus, and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106408001A (en) * | 2016-08-26 | 2017-02-15 | 西安电子科技大学 | Rapid area-of-interest detection method based on depth kernelized hashing |
| CN107563123A (en) * | 2017-09-27 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | Method and apparatus for marking medical image |
| US20180012359A1 (en) * | 2016-07-06 | 2018-01-11 | Marinko Venci Sarunic | Systems and Methods for Automated Image Classification and Segmentation |
| CN108108757A (en) * | 2017-12-18 | 2018-06-01 | 深圳市唯特视科技有限公司 | A kind of diabetic foot ulcers sorting technique based on convolutional neural networks |
| US20180322660A1 (en) * | 2017-05-02 | 2018-11-08 | Techcyte, Inc. | Machine learning classification and training for digital microscopy images |
| CN108960289A (en) * | 2018-06-08 | 2018-12-07 | 清华大学 | Medical imaging sorter and method |
| CN109657082A (en) * | 2018-08-28 | 2019-04-19 | 武汉大学 | Remote sensing images multi-tag search method and system based on full convolutional neural networks |
-
2019
- 2019-05-17 CN CN201910411362.2A patent/CN110245657B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180012359A1 (en) * | 2016-07-06 | 2018-01-11 | Marinko Venci Sarunic | Systems and Methods for Automated Image Classification and Segmentation |
| CN106408001A (en) * | 2016-08-26 | 2017-02-15 | 西安电子科技大学 | Rapid area-of-interest detection method based on depth kernelized hashing |
| US20180322660A1 (en) * | 2017-05-02 | 2018-11-08 | Techcyte, Inc. | Machine learning classification and training for digital microscopy images |
| CN107563123A (en) * | 2017-09-27 | 2018-01-09 | 百度在线网络技术（北京）有限公司 | Method and apparatus for marking medical image |
| CN108108757A (en) * | 2017-12-18 | 2018-06-01 | 深圳市唯特视科技有限公司 | A kind of diabetic foot ulcers sorting technique based on convolutional neural networks |
| CN108960289A (en) * | 2018-06-08 | 2018-12-07 | 清华大学 | Medical imaging sorter and method |
| CN109657082A (en) * | 2018-08-28 | 2019-04-19 | 武汉大学 | Remote sensing images multi-tag search method and system based on full convolutional neural networks |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11468565B2 (en) | 2019-09-30 | 2022-10-11 | Institute Of Computing Technology, Chinese Academy Of Sciences | TMB classification method and system and TMB analysis device based on pathological image |
| CN110866893B (en) * | 2019-09-30 | 2021-04-06 | 中国科学院计算技术研究所 | Pathological image-based TMB classification method and system and TMB analysis device |
| CN110866893A (en) * | 2019-09-30 | 2020-03-06 | 中国科学院计算技术研究所 | Pathological image-based TMB classification method and system and TMB analysis device |
| CN111091539A (en) * | 2019-12-09 | 2020-05-01 | 上海联影智能医疗科技有限公司 | Network model training method, medical image processing method, device, medium and equipment |
| CN111091539B (en) * | 2019-12-09 | 2024-03-26 | 上海联影智能医疗科技有限公司 | Network model training and medical image processing methods, devices, mediums and equipment |
| CN111368923A (en) * | 2020-03-05 | 2020-07-03 | 上海商汤智能科技有限公司 | Neural network training method and device, electronic equipment and storage medium |
| CN111368923B (en) * | 2020-03-05 | 2023-12-19 | 上海商汤智能科技有限公司 | Neural network training method and device, electronic equipment and storage medium |
| CN111553361A (en) * | 2020-03-19 | 2020-08-18 | 四川大学华西医院 | Pathological section label identification method |
| CN111553361B (en) * | 2020-03-19 | 2022-11-01 | 四川大学华西医院 | Pathological section label identification method |
| TWI779284B (en) * | 2020-05-06 | 2022-10-01 | 商之器科技股份有限公司 | Device for marking image data |
| CN114419314A (en) * | 2020-10-09 | 2022-04-29 | 通用电气精准医疗有限责任公司 | System and method for automatically labeling medical images |
| CN112287843A (en) * | 2020-10-30 | 2021-01-29 | 西安工程大学 | Doctor and nurse interaction gesture recognition method in operation process |
| CN112614119A (en) * | 2020-12-28 | 2021-04-06 | 上海市精神卫生中心（上海市心理咨询培训中心） | Medical image region-of-interest visualization method, device, storage medium and equipment |
| CN112614119B (en) * | 2020-12-28 | 2024-04-12 | 上海市精神卫生中心（上海市心理咨询培训中心） | Medical image region of interest visualization method, device, storage medium and equipment |
| WO2024016691A1 (en) * | 2022-07-22 | 2024-01-25 | 华为云计算技术有限公司 | Image retrieval method and apparatus, model training method and apparatus, and storage medium |
| CN117115183A (en) * | 2023-09-01 | 2023-11-24 | 北京透彻未来科技有限公司 | Interception method and system based on digital pathological image visual area |
| CN117115183B (en) * | 2023-09-01 | 2024-02-09 | 北京透彻未来科技有限公司 | Interception method and system based on digital pathological image visual area |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110245657B (en) | 2021-08-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110245657A (en) | Pathological image similarity detection method and detection device | |
| CN106056595B (en) | Based on the pernicious assistant diagnosis system of depth convolutional neural networks automatic identification Benign Thyroid Nodules | |
| CN106815481B (en) | Lifetime prediction method and device based on image omics | |
| CN108257135A (en) | The assistant diagnosis system of medical image features is understood based on deep learning method | |
| CN111931811B (en) | Calculation method based on super-pixel image similarity | |
| Zhang et al. | Automated semantic segmentation of red blood cells for sickle cell disease | |
| CN103678504B (en) | Similarity-based breast image matching image searching method and system | |
| Pan et al. | Cell detection in pathology and microscopy images with multi-scale fully convolutional neural networks | |
| WO2021051875A1 (en) | Cell classification method and apparatus, medium and electronic device | |
| WO2023045231A1 (en) | Method and apparatus for facial nerve segmentation by decoupling and divide-and-conquer | |
| CN110059656B (en) | Method and system for classifying white blood cells based on convolution countermeasure generation neural network | |
| US11176412B2 (en) | Systems and methods for encoding image features of high-resolution digital images of biological specimens | |
| CN112347977B (en) | Automatic detection method, storage medium and device for induced pluripotent stem cells | |
| CN108664986B (en) | Based on lpNorm regularized multi-task learning image classification method and system | |
| CN112263217B (en) | Improved convolutional neural network-based non-melanoma skin cancer pathological image lesion area detection method | |
| CN112819747A (en) | Method for automatically diagnosing benign and malignant nodules based on lung tomography image | |
| CN115546605A (en) | Training method and device based on image labeling and segmentation model | |
| CN111899259A (en) | Prostate cancer tissue microarray classification method based on convolutional neural network | |
| CN114399634B (en) | Three-dimensional image classification method, system, equipment and medium based on weak supervision learning | |
| CN113288157A (en) | Arrhythmia classification method based on depth separable convolution and improved loss function | |
| CN112330645A (en) | Glioma grading method and device based on attention mechanism | |
| Shi et al. | High-throughput fat quantifications of hematoxylin-eosin stained liver histopathological images based on pixel-wise clustering | |
| Zeng et al. | Unsupervised Skin Lesion Segmentation via Structural Entropy Minimization on Multi-Scale Superpixel Graphs | |
| Kumar et al. | Prediction of Lung Cancer Using Machine Learning Technique: A Survey | |
| Yao et al. | Brain hematoma segmentation using active learning and an active contour model |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |