CN108898160A - Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features - Google Patents
Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features Download PDFInfo
- Publication number
- CN108898160A CN108898160A CN201810555876.0A CN201810555876A CN108898160A CN 108898160 A CN108898160 A CN 108898160A CN 201810555876 A CN201810555876 A CN 201810555876A CN 108898160 A CN108898160 A CN 108898160A
- Authority
- CN
- China
- Prior art keywords
- cnn
- model
- image
- breast cancer
- fusion features
- 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
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to CNN and image classification identification technology fields, more particularly to breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features.The present invention proposes that the CNN model merged by construction feature judges breast cancer tissue's pathology grade of molybdenum target image, gray feature, textural characteristics and the wavelet character extracted using molybdenum target tumor region, Feature Selection is carried out by LASSO logistic regression model, select the feature big with breast cancer tissue pathology rank correlation, Fusion Features are carried out in the newly added full articulamentum of network by the high-level semantics features for extracting CNN and the image group feature filtered out again, and the CNN model that fitting obtains Fusion Features is used to identify breast cancer tissue's pathology grade.The present invention can the breast molybdenum target image directly to patient scan carry out breast cancer tissue's pathology grade locating for analytical judgment patient, further shortened while guaranteeing discrimination precision differentiate the time.
Description
Technical field
The present invention relates to CNN and image classification identification technology fields, more particularly to are based on CNN and image group Fusion Features
Breast cancer tissue's Pathologic Grading method.
Background technique
The breast cancer cancer common as women is second disease for being easiest to cause women die.Global breast cancer
Disease incidence is in rising trend always since the late 1970s, and because the patient of breast cancer deaths is not within minority.Mammary gland molybdenum
Target X-ray photographic Examined effect is current preferred and most easy, the most reliable non-invasive detection means for differentiating mammary gland disease, and
High resolution facilitates early detection breast cancer.
In recent years, with the development of big data and high-performance calculation, CNN (convolutional neural networks) is in computer vision field
The achievement for achieving conspicuousness, the discrimination on scene image classification have been more than that mankind's identification is horizontal.CNN passes through multilayer convolution
With pondization extract characteristics of image, then by back-propagation algorithm progress parameter update, change previous artificial design features by
The limitation of people's experience.
The histopathology histological grading (SBR classification) of breast cancer mainly passes through the mitotic index of cancer cell, mammary gland gland
The morphological image characteristic of three aspects of atypia of the difference and cancer cell core of pipe, which is combined, to be assessed, the tissue of breast cancer
The prognosis of Pathologic Grading and patient have important relationship, and within same clinical stages, 5 years survival rates of patient are with tissue
The raising of Pathologic Grading and decline.And breast cancer SBR classification is differentiated mainly by observing patient under the microscope
The Carcinoma cell differentiation situation of pathological section, doctor still directly cannot carry out classification differentiation from conventional molybdenum target image at present.
Summary of the invention
In view of the above-mentioned problems, the invention proposes breast cancer tissue's pathology based on CNN and image group Fusion Features
Stage division can be analyzed by the breast molybdenum target image directly to patient, by the way that the image group of engineer is special
The image high-level semantics features that the CNN that seeks peace is automatically extracted carry out Fusion Features on newly added full articulamentum, and training characteristics are melted
CNN model after conjunction is to obtain breast cancer tissue's pathology grade locating for patient, for the differentiation of further disease and prognosis
Analysis provides foundation.
To achieve the goals above, the present invention uses following technical scheme：
Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features, includes the following steps：
Step 1：Breast molybdenum target image tumor region is extracted, on the molybdenum target tumor region of extraction carry out gray scale,
The calculating of texture and wavelet character extracts 180 dimension image group feature vectors by above-mentioned calculating altogether；By the breast molybdenum target of extraction
Image tumor region is fabricated to the tumor of breast region molybdenum target image pattern of same size, by image pattern be divided into training set,
Verifying collection and test set；
Step 2：Image group feature vector is tieed up to the 180 of extraction, is carried out using LASSO logistic regression model special
Sign screening, using the image group feature after screening to carry out Fusion Features；
Step 3：Transfer learning, training CNN hierarchy model, in CNN hierarchy model are carried out using the CNN model of pre-training
New full articulamentum is added on original base, on new full articulamentum by before the full articulamentum of CNN hierarchy model output and
Image group feature after screening carries out Fusion Features, and carries out retraining on the basis of CNN hierarchy model parameter, and update is melted
CNN hierarchy model parameter after conjunction, according to model verifying collection on grading effect to fused CNN hierarchy model parameter into
Row adjustment, obtains the CNN model of Fusion Features, for carrying out breast cancer tissue's Pathologic Grading to breast molybdenum target image.
Further, further include after the step 3：
Model Grading accuracy rate is verified to the CNN model for obtaining Fusion Features using test set.
Further, the step 1 includes：
Step 1.1：ROI extraction is carried out to breast molybdenum target image tumor region, ROI image is obtained, calculates the 14 of ROI image
A gray feature, 22 textural characteristics and 144 wavelet characters extract 180 dimension image group feature vectors altogether；
Step 1.2：Expand the scale of ROI image by data enhancement methods；
Step 1.3：ROI image after data scale is expanded zooms to same size uniformly to adapt to the defeated of CNN model
Enter requirement.
Further, the step 3 includes：
Step 3.1：Using tumor of breast region molybdenum target image pattern in training set as the input of CNN model, ImageNet from
Transfer learning, training CNN hierarchy model are carried out on right image data set on the CNN model of pre-training；
Step 3.2：A new full articulamentum is added on the original base of CNN hierarchy model, it will on new full articulamentum
Tumor of breast region molybdenum target image high-level semantics features before the full articulamentum of CNN hierarchy model export and use LASSO
Logistic regression model screening image group feature progress Fusion Features, and on the basis of CNN hierarchy model parameter into
Row retraining updates fused CNN hierarchy model parameter, according to grading effect of the model on verifying collection to fused
CNN hierarchy model parameter is adjusted, and obtains the CNN model of Fusion Features.
Compared with prior art, the device have the advantages that：
The present invention proposes that the CNN model merged by construction feature judges breast cancer tissue's pathology grade of molybdenum target image, benefit
With molybdenum target tumor region extract gray feature, textural characteristics and wavelet character, by LASSO logistic regression model into
Row Feature Selection selects the feature big with breast cancer tissue pathology rank correlation, then the high-level semantic by extracting CNN
Feature and the image group feature filtered out are in the newly added full articulamentum progress Fusion Features of network, and fitting obtains feature and melts
The CNN model of conjunction is used to identify breast cancer tissue's pathology grade.The present invention can directly to patient scan breast molybdenum target figure
As carrying out breast cancer tissue's pathology grade locating for analytical judgment patient, further shortened while guaranteeing discrimination precision
Differentiate the time.
Detailed description of the invention
Fig. 1 is the Pathologic Grading side of breast cancer tissue based on CNN and image group Fusion Features of the embodiment of the present invention
The basic flow chart of method.
Fig. 2 is breast cancer tissue's pathology credit based on CNN and image group Fusion Features of another embodiment of the present invention
The basic flow chart of grade method.
Fig. 3 is the Pathologic Grading side of breast cancer tissue based on CNN and image group Fusion Features of the embodiment of the present invention
The different molybdenum target images thrown according to position of method.
Specific embodiment
With reference to the accompanying drawing with specific embodiment the present invention will be further explained explanation：
Embodiment one：
As shown in Figure 1, a kind of breast cancer tissue's Pathologic Grading based on CNN and image group Fusion Features of the invention
Method includes the following steps：
Step S101：Breast molybdenum target image tumor region is extracted, ash is carried out on the molybdenum target tumor region of extraction
The calculating of degree, texture and wavelet character, extracts 180 dimension image group feature vectors by above-mentioned calculating altogether；By the mammary gland of extraction
Molybdenum target image tumor region is fabricated to the tumor of breast region molybdenum target image pattern of same size, and image pattern is divided into training
Collection, verifying collection and test set.
Step S102：Image group feature vector is tieed up to the 180 of extraction, is carried out using LASSO logistic regression model
Feature Selection, using the image group feature after screening to carry out Fusion Features.
Step S103：Transfer learning is carried out using the CNN model of pre-training, training CNN hierarchy model is classified mould in CNN
New full articulamentum is added on the original base of type, it will be defeated before the full articulamentum of CNN hierarchy model on new full articulamentum
Image group feature out and after screening carries out Fusion Features, and carries out retraining on the basis of CNN hierarchy model parameter, obtains
To fused CNN model, fused CNN hierarchy model parameter is adjusted according to grading effect of the model on verifying collection
It is whole, the CNN model of Fusion Features is obtained, for carrying out breast cancer tissue's Pathologic Grading to breast molybdenum target image.
Embodiment two：
As shown in Fig. 2, another breast cancer tissue's pathology credit based on CNN and image group Fusion Features of the invention
Grade method, includes the following steps：
Step S201：Breast molybdenum target image tumor region is extracted, ash is carried out on the molybdenum target tumor region of extraction
The calculating of degree, texture and wavelet character, extracts 180 dimension image group feature vectors by above-mentioned calculating altogether；By the mammary gland of extraction
Molybdenum target image tumor region is fabricated to the tumor of breast region molybdenum target image pattern of same size, and image pattern is divided into training
Collection, verifying collection and test set.
The step S201 includes：
Step S2011：ROI extraction is carried out to breast molybdenum target image tumor region, ROI image is obtained, calculates ROI image
14 gray features, 22 textural characteristics and 144 wavelet characters extract 180 dimension image group feature vectors altogether；
The gray feature is gray scale maximum value, minimum value, mean value, intermediate value, variance, kurtosis, energy, entropy, absolute variance
Mean value, skewness, standard deviation, the uniformity, gray scale codomain, root mean square totally 14 features (referring to Aerts H J W L,
Velazquez E R,Leijenaar R T H,et al.Decoding tumour phenotype by noninvasive
imaging using a quantitative radiomics approach[J].Nature communications,
2014,5:4006)；
The textural characteristics are 9 dimensional features derived based on gray level co-occurrence matrixes, i.e. energy, contrast, entropy, homogeneity
Property, correlation, variance and average, otherness, auto-correlation be (referring to Weszka J S, Dyer C R, Rosenfeld A.A
comparative study of texture measures for terrain classification[J].IEEE
transactions on Systems,Man,and Cybernetics,1976(4):269-285) and based on gray scale distance of swimming square
13 dimensional features that battle array derives, i.e. Short Run Emphasis, Long Run Emphasis, Gray-Level
Nonuniformity、Run-Length Nonuniformity、Run Percentage、Low Gray-Level Run
Emphasis、High Gray-Level Run Emphasis、Short Run Low Gray-Level Emphasis、Short
Run High Gray-Level Emphasis、Long Run Low Gray-Level Emphasis、Long Run High
Gray-Level Emphasis, Gray-Level Variance, Run-Length Variance are (referring to Galloway M
M.Texture analysis using grey level run lengths[J].NASA STI/Recon Technical
Report N,1974,75；Chu A,Sehgal C M,Greenleaf J F.Use of gray value
distribution of run lengths for texture analysis[J].Pattern Recognition
Letters,1990,11(6):Totally 22 dimensional feature 415-419)；
The wavelet character is to calculate separately gray feature and textural characteristics in 4 Wavelet Components, totally 144 features.
Step S2012：Expand the scale of ROI image by data enhancement methods；As an embodiment, Ke Yitong
The data enhancement methods for crossing random translation, rotation, overturning and multiple dimensioned scaling expand the scale of ROI image；
Step S2013：ROI image after data scale is expanded zooms to same size uniformly to adapt to CNN model
Input requirements.
Step S202：Image group feature vector is tieed up to the 180 of extraction, is carried out using LASSO logistic regression model
Feature Selection selects the feature big with breast cancer tissue pathology rank correlation, using the image group feature after screening with
Carry out Fusion Features.
It is that L1 regularization term is added on the basis of least square fitting to improve the essence of linear regression model (LRM) that LASSO, which is returned,
Degree, its penalty is the absolute value of regression coefficient, this can make some parameter estimation results be equal to zero, therefore facilitate feature
Selection.Histopathology histological grading is the classification problem of a binary, and Logistic regression analysis is binary classification or one-to-many
Classify common generalized linear model, the response of simple linear regression is normalized to 0 and 1 by it, therefore LASSO can be returned mould
Linear regression in type is replaced by logistic and returns to select the feature of binary classification.LASSO logistic regression optimization
Objective function is as follows：
Wherein, n is the number of sample, XiIt is the initial data of m × n size, i.e., each sample has m feature, yiIt is every
The corresponding response of a sample, ω is linear regression coeffficient, and b is the cutoff value of linear regression, and λ is dilute for controlling regression coefficient
Dredge the non-negative regularization parameter of degree.The image group feature input LASSO logistic regression model of extraction can be subjected to shadow
As group learns Feature Selection.
Step S203：Transfer learning is carried out using the CNN model of pre-training, training CNN hierarchy model is classified mould in CNN
New full articulamentum is added on the original base of type, it will be defeated before the full articulamentum of CNN hierarchy model on new full articulamentum
Image group feature out and after screening carries out Fusion Features, and carries out retraining on the basis of CNN hierarchy model parameter, more
New fused CNN hierarchy model parameter joins fused CNN hierarchy model according to grading effect of the model on verifying collection
Number is adjusted, and obtains the CNN model of Fusion Features, for carrying out breast cancer tissue's Pathologic Grading to breast molybdenum target image.
The step S203 includes：
Step S2031：Using tumor of breast region molybdenum target image pattern in training set as the input of CNN model,
Transfer learning, training CNN hierarchy model are carried out on ImageNet natural image data set on the CNN model of pre-training；
Step S2032：A new full articulamentum is added on the original base of CNN hierarchy model, in new full connection
By the output of tumor of breast region molybdenum target image high-level semantics features and use before the full articulamentum of CNN hierarchy model on layer
The image group feature of LASSO logistic regression model screening carries out Fusion Features, and in the base of CNN hierarchy model parameter
Retraining is carried out on plinth, updates fused CNN hierarchy model parameter, according to grading effect of the model on verifying collection to fusion
CNN hierarchy model parameter afterwards is adjusted, and obtains the CNN model of Fusion Features.
Step S204：Model Grading accuracy rate is verified using CNN model of the test set to obtained Fusion Features.
As an embodiment, the breast molybdenum target image data set used shares 204 cases, and each case includes
Axle position (craniocaudal, CC) image and lateral oblique position (mediolateral oblique, MLO) image, as shown in figure 3, Fig. 3
In (a) be partially axle position molybdenum target image, (b) is partially deviational survey position molybdenum target image in Fig. 3.The storage of molybdenum target image is using standard
DICOM format, resolution ratio (wide × high) have 3328 × 4096 and 2560 × 3,328 two kinds.Tumor area in all molybdenum target images
Domain is all that the radiologist through hospital's profession delineates, and all cases are equipped with the Accurate Diagnosis result of hospital pathology department
To determine its pathology grade.By to different breast cancer molybdenum target imaged tissue Pathologic Grading algorithms collection mammary gland molybdenum
Target image data set is tested, and quantitative assessment is carried out to classification performance using classification accuracy and AUC value, as a result such as 1 institute of table
Show.The embodiment of the present invention is compared based on CNN with breast cancer tissue's Pathologic Grading method of image group Fusion Features
GoogLeNet is (referring to Szegedy C, Liu W, Jia Y, et al.Going deeper with convolutions [C]
.IEEE Conference on Computer Vision and Pattern Recognition,2015:It is 1-9) and traditional
The classifying quality of random forest grader, which has, to be obviously improved, and classification accuracy reaches 0.7500, and AUC value reaches 0.8051.
1 breast cancer molybdenum target image Pathologic Grading algorithm classification performance of table
| Sorting algorithm | Classification accuracy | AUC |
| GoogLeNet | 0.7031 | 0.7049 |
| Random forest | 0.6029 | 0.6618 |
| Feature Fusion Algorithm | 0.7500 | 0.8051 |
Illustrated above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (4)
1. breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features, which is characterized in that including following
Step：
Step 1：Breast molybdenum target image tumor region is extracted, gray scale, texture are carried out on the molybdenum target tumor region of extraction
With the calculating of wavelet character, 180 dimension image group feature vectors are extracted by above-mentioned calculating altogether；By the breast molybdenum target image of extraction
Tumor region is fabricated to the tumor of breast region molybdenum target image pattern of same size, and image pattern is divided into training set, verifying
Collection and test set；
Step 2：Image group feature vector is tieed up to the 180 of extraction, feature sieve is carried out using LASSO logistic regression model
Choosing, using the image group feature after screening to carry out Fusion Features；
Step 3：Transfer learning, training CNN hierarchy model, in the original of CNN hierarchy model are carried out using the CNN model of pre-training
On the basis of add new full articulamentum, by the output and screening before the full articulamentum of CNN hierarchy model on new full articulamentum
Image group feature afterwards carries out Fusion Features, and retraining is carried out on the basis of CNN hierarchy model parameter, after updating fusion
CNN hierarchy model parameter, according to model verifying collection on grading effect fused CNN hierarchy model parameter is adjusted
It is whole, the CNN model of Fusion Features is obtained, for carrying out breast cancer tissue's Pathologic Grading to breast molybdenum target image.
2. breast cancer tissue's Pathologic Grading method according to claim 1 based on CNN and image group Fusion Features,
It is characterized in that, further including after the step 3：
Model Grading accuracy rate is verified to the CNN model for obtaining Fusion Features using test set.
3. breast cancer tissue's Pathologic Grading method according to claim 1 based on CNN and image group Fusion Features,
It is characterized in that, the step 1 includes：
Step 1.1：ROI extraction is carried out to breast molybdenum target image tumor region, ROI image is obtained, calculates 14 ashes of ROI image
Feature, 22 textural characteristics and 144 wavelet characters are spent, extract 180 dimension image group feature vectors altogether；
Step 1.2：Expand the scale of ROI image by data enhancement methods；
Step 1.3：ROI image after data scale is expanded uniformly is zoomed to same size and is wanted with the input for adapting to CNN model
It asks.
4. breast cancer tissue's Pathologic Grading method according to claim 1 based on CNN and image group Fusion Features,
It is characterized in that, the step 3 includes：
Step 3.1：Using tumor of breast region molybdenum target image pattern in training set as the input of CNN model, ImageNet from
Transfer learning, training CNN hierarchy model are carried out on right image data set on the CNN model of pre-training；
Step 3.2：A new full articulamentum is added on the original base of CNN hierarchy model, it will on new full articulamentum
The output of tumor of breast region molybdenum target image high-level semantics features and use before the full articulamentum of CNN hierarchy model
The image group feature of LASSOlogistic regression model screening carries out Fusion Features, and on the basis of CNN hierarchy model parameter
Upper carry out retraining, updates fused CNN hierarchy model parameter, according to model after the grading effect on verifying collection is to fusion
CNN hierarchy model parameter be adjusted, obtain the CNN model of Fusion Features.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810555876.0A CN108898160B (en) | 2018-06-01 | 2018-06-01 | Breast cancer histopathology grading method based on CNN and imaging omics feature fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810555876.0A CN108898160B (en) | 2018-06-01 | 2018-06-01 | Breast cancer histopathology grading method based on CNN and imaging omics feature fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108898160A true CN108898160A (en) | 2018-11-27 |
| CN108898160B CN108898160B (en) | 2022-04-08 |
Family
ID=64343881
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810555876.0A Active CN108898160B (en) | 2018-06-01 | 2018-06-01 | Breast cancer histopathology grading method based on CNN and imaging omics feature fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108898160B (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583447A (en) * | 2018-12-11 | 2019-04-05 | 山东省肿瘤防治研究院（山东省肿瘤医院） | A kind of image group credit analysis model building method and analysis method |
| CN109816632A (en) * | 2018-12-25 | 2019-05-28 | 东软集团股份有限公司 | Brain image processing method, device, readable storage medium storing program for executing and electronic equipment |
| CN109919187A (en) * | 2019-01-28 | 2019-06-21 | 浙江工商大学 | With bagging fine tuning CNN come the method for Thyroid Follicular picture of classifying |
| CN109948667A (en) * | 2019-03-01 | 2019-06-28 | 桂林电子科技大学 | Image classification method and device for the prediction of correct neck cancer far-end transfer |
| CN109993740A (en) * | 2019-04-02 | 2019-07-09 | 广州中医药大学(广州中医药研究院) | A kind of cell cancer image grading method and device |
| CN110175978A (en) * | 2019-04-02 | 2019-08-27 | 南方医科大学南方医院 | A kind of liver cancer image group data processing method, system, device and storage medium |
| CN110533683A (en) * | 2019-08-30 | 2019-12-03 | 东南大学 | A kind of image group analysis method merging traditional characteristic and depth characteristic |
| CN110689960A (en) * | 2019-09-24 | 2020-01-14 | 广州大学 | Intelligent breast cancer distinguishing method, device, equipment and storage medium |
| CN111242174A (en) * | 2019-12-31 | 2020-06-05 | 浙江大学 | Liver cancer image feature extraction and pathological classification method and device based on imaging omics |
| CN111415741A (en) * | 2020-03-05 | 2020-07-14 | 北京深睿博联科技有限责任公司 | Breast X-ray image classification model training method based on implicit appearance learning |
| CN111583320A (en) * | 2020-03-17 | 2020-08-25 | 哈尔滨医科大学 | Breast cancer ultrasonic image typing method and system fusing deep convolutional network and image omics characteristics and storage medium |
| CN111739033A (en) * | 2020-06-22 | 2020-10-02 | 中山大学肿瘤防治中心（中山大学附属肿瘤医院、中山大学肿瘤研究所） | Method for establishing breast molybdenum target and MR image omics model based on machine learning |
| CN111914952A (en) * | 2020-08-21 | 2020-11-10 | 山东省医学影像学研究所 | AD characteristic parameter screening method and system based on deep neural network |
| CN112365440A (en) * | 2020-10-10 | 2021-02-12 | 天津大学 | Yoov 3-based mastopathy image mitosis detection device and method |
| CN113436158A (en) * | 2021-06-21 | 2021-09-24 | 遂宁市中心医院 | Liver lump auxiliary identification method based on deep learning |
| CN114171197A (en) * | 2021-11-12 | 2022-03-11 | 东莞市人民医院 | Method and related equipment for predicting HER2 state of breast cancer |
| CN114360718A (en) * | 2022-03-10 | 2022-04-15 | 之江实验室 | Feature fitting-based PET/CT automatic lung cancer diagnosis and classification system and construction method |
| CN114511564A (en) * | 2022-04-19 | 2022-05-17 | 天津市肿瘤医院（天津医科大学肿瘤医院） | Image analysis method for breast cancer residual tumor load based on DCE-MRI |
| WO2023115333A1 (en) * | 2021-12-21 | 2023-06-29 | 深圳先进技术研究院 | Expert knowledge constraint-based deep learning method for brain glioma grading |
| CN116883995A (en) * | 2023-07-07 | 2023-10-13 | 广东食品药品职业学院 | Identification system of breast cancer molecular subtype |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096654A (en) * | 2016-06-13 | 2016-11-09 | 南京信息工程大学 | A kind of cell atypia automatic grading method tactful based on degree of depth study and combination |
| CN106326931A (en) * | 2016-08-25 | 2017-01-11 | 南京信息工程大学 | Mammary gland molybdenum target image automatic classification method based on deep learning |
| CN106599883A (en) * | 2017-03-08 | 2017-04-26 | 王华锋 | Face recognition method capable of extracting multi-level image semantics based on CNN (convolutional neural network) |
| CN106682435A (en) * | 2016-12-31 | 2017-05-17 | 西安百利信息科技有限公司 | System and method for automatically detecting lesions in medical image through multi-model fusion |
| CN106683081A (en) * | 2016-12-17 | 2017-05-17 | 复旦大学 | Brain glioma molecular marker nondestructive prediction method and prediction system based on radiomics |
| US20170140253A1 (en) * | 2015-11-12 | 2017-05-18 | Xerox Corporation | Multi-layer fusion in a convolutional neural network for image classification |
| CN106780448A (en) * | 2016-12-05 | 2017-05-31 | 清华大学 | A kind of pernicious sorting technique of ultrasonic Benign Thyroid Nodules based on transfer learning Yu Fusion Features |
| CN107330263A (en) * | 2017-06-26 | 2017-11-07 | 成都知识视觉科技有限公司 | A kind of method of area of computer aided breast invasive ductal carcinoma histological grading |
| CN108062753A (en) * | 2017-12-29 | 2018-05-22 | 重庆理工大学 | The adaptive brain tumor semantic segmentation method in unsupervised domain based on depth confrontation study |
-
2018
- 2018-06-01 CN CN201810555876.0A patent/CN108898160B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170140253A1 (en) * | 2015-11-12 | 2017-05-18 | Xerox Corporation | Multi-layer fusion in a convolutional neural network for image classification |
| CN106096654A (en) * | 2016-06-13 | 2016-11-09 | 南京信息工程大学 | A kind of cell atypia automatic grading method tactful based on degree of depth study and combination |
| CN106326931A (en) * | 2016-08-25 | 2017-01-11 | 南京信息工程大学 | Mammary gland molybdenum target image automatic classification method based on deep learning |
| CN106780448A (en) * | 2016-12-05 | 2017-05-31 | 清华大学 | A kind of pernicious sorting technique of ultrasonic Benign Thyroid Nodules based on transfer learning Yu Fusion Features |
| CN106683081A (en) * | 2016-12-17 | 2017-05-17 | 复旦大学 | Brain glioma molecular marker nondestructive prediction method and prediction system based on radiomics |
| CN106682435A (en) * | 2016-12-31 | 2017-05-17 | 西安百利信息科技有限公司 | System and method for automatically detecting lesions in medical image through multi-model fusion |
| CN106599883A (en) * | 2017-03-08 | 2017-04-26 | 王华锋 | Face recognition method capable of extracting multi-level image semantics based on CNN (convolutional neural network) |
| CN107330263A (en) * | 2017-06-26 | 2017-11-07 | 成都知识视觉科技有限公司 | A kind of method of area of computer aided breast invasive ductal carcinoma histological grading |
| CN108062753A (en) * | 2017-12-29 | 2018-05-22 | 重庆理工大学 | The adaptive brain tumor semantic segmentation method in unsupervised domain based on depth confrontation study |
Non-Patent Citations (5)
| Title |
|---|
| FABIO ALEXANDRE SPANHOL 等: "Breast cancer histopathological image classification using Convolutional Neural Networks", 《2016 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS》 * |
| HAIBO WANG 等: "Mitosis detection in breast cancer pathology images by combining handcrafted and convolutional neural network features", 《JOURNAL OF MEDICAL IMAGING》 * |
| YAN-QI HUANG 等: "Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer", 《JOURNAL OF CLINICAL ONCOLOGY》 * |
| 周超 等: "基于深度卷积网络和结合策略的乳腺组织病理图像细胞核异型性自动评分", 《中国生物医学工程学报》 * |
| 龚磊 等: "基于多特征描述的乳腺癌肿瘤病理自动分级", 《计算机应用》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583447A (en) * | 2018-12-11 | 2019-04-05 | 山东省肿瘤防治研究院（山东省肿瘤医院） | A kind of image group credit analysis model building method and analysis method |
| CN109816632A (en) * | 2018-12-25 | 2019-05-28 | 东软集团股份有限公司 | Brain image processing method, device, readable storage medium storing program for executing and electronic equipment |
| CN109919187A (en) * | 2019-01-28 | 2019-06-21 | 浙江工商大学 | With bagging fine tuning CNN come the method for Thyroid Follicular picture of classifying |
| CN109919187B (en) * | 2019-01-28 | 2021-02-12 | 浙江工商大学 | Method for classifying thyroid follicular picture by using bagging fine tuning CNN |
| CN109948667A (en) * | 2019-03-01 | 2019-06-28 | 桂林电子科技大学 | Image classification method and device for the prediction of correct neck cancer far-end transfer |
| CN109993740A (en) * | 2019-04-02 | 2019-07-09 | 广州中医药大学(广州中医药研究院) | A kind of cell cancer image grading method and device |
| CN110175978A (en) * | 2019-04-02 | 2019-08-27 | 南方医科大学南方医院 | A kind of liver cancer image group data processing method, system, device and storage medium |
| CN110533683A (en) * | 2019-08-30 | 2019-12-03 | 东南大学 | A kind of image group analysis method merging traditional characteristic and depth characteristic |
| CN110689960A (en) * | 2019-09-24 | 2020-01-14 | 广州大学 | Intelligent breast cancer distinguishing method, device, equipment and storage medium |
| CN111242174A (en) * | 2019-12-31 | 2020-06-05 | 浙江大学 | Liver cancer image feature extraction and pathological classification method and device based on imaging omics |
| CN111415741A (en) * | 2020-03-05 | 2020-07-14 | 北京深睿博联科技有限责任公司 | Breast X-ray image classification model training method based on implicit appearance learning |
| CN111415741B (en) * | 2020-03-05 | 2023-09-26 | 北京深睿博联科技有限责任公司 | Mammary gland X-ray image classification model training method based on implicit apparent learning |
| CN111583320A (en) * | 2020-03-17 | 2020-08-25 | 哈尔滨医科大学 | Breast cancer ultrasonic image typing method and system fusing deep convolutional network and image omics characteristics and storage medium |
| CN111739033A (en) * | 2020-06-22 | 2020-10-02 | 中山大学肿瘤防治中心（中山大学附属肿瘤医院、中山大学肿瘤研究所） | Method for establishing breast molybdenum target and MR image omics model based on machine learning |
| CN111914952A (en) * | 2020-08-21 | 2020-11-10 | 山东省医学影像学研究所 | AD characteristic parameter screening method and system based on deep neural network |
| CN111914952B (en) * | 2020-08-21 | 2024-03-08 | 山东第一医科大学附属省立医院(山东省立医院) | AD characteristic parameter screening method and system based on deep neural network |
| CN112365440A (en) * | 2020-10-10 | 2021-02-12 | 天津大学 | Yoov 3-based mastopathy image mitosis detection device and method |
| CN113436158A (en) * | 2021-06-21 | 2021-09-24 | 遂宁市中心医院 | Liver lump auxiliary identification method based on deep learning |
| CN114171197A (en) * | 2021-11-12 | 2022-03-11 | 东莞市人民医院 | Method and related equipment for predicting HER2 state of breast cancer |
| WO2023115333A1 (en) * | 2021-12-21 | 2023-06-29 | 深圳先进技术研究院 | Expert knowledge constraint-based deep learning method for brain glioma grading |
| CN114360718A (en) * | 2022-03-10 | 2022-04-15 | 之江实验室 | Feature fitting-based PET/CT automatic lung cancer diagnosis and classification system and construction method |
| CN114360718B (en) * | 2022-03-10 | 2022-07-12 | 之江实验室 | Feature fitting-based PET/CT automatic lung cancer diagnosis and classification system and construction method |
| CN114511564A (en) * | 2022-04-19 | 2022-05-17 | 天津市肿瘤医院（天津医科大学肿瘤医院） | Image analysis method for breast cancer residual tumor load based on DCE-MRI |
| CN116883995A (en) * | 2023-07-07 | 2023-10-13 | 广东食品药品职业学院 | Identification system of breast cancer molecular subtype |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108898160B (en) | 2022-04-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108898160A (en) | Breast cancer tissue's Pathologic Grading method based on CNN and image group Fusion Features | |
| Sharma et al. | [Retracted] Enhanced Watershed Segmentation Algorithm‐Based Modified ResNet50 Model for Brain Tumor Detection | |
| Bai et al. | Liver tumor segmentation based on multi-scale candidate generation and fractal residual network | |
| WO2021203795A1 (en) | Pancreas ct automatic segmentation method based on saliency dense connection expansion convolutional network | |
| CN109191471A (en) | Based on the pancreatic cell image partition method for improving U-Net network | |
| CN109034045A (en) | A kind of leucocyte automatic identifying method based on convolutional neural networks | |
| US20150071541A1 (en) | Automated method for measuring, classifying, and matching the dynamics and information passing of single objects within one or more images | |
| CN106815481A (en) | A kind of life cycle Forecasting Methodology and device based on image group | |
| CN112862824A (en) | Novel coronavirus pneumonia focus detection method, system, device and storage medium | |
| Sarosa et al. | Mammogram breast cancer classification using gray-level co-occurrence matrix and support vector machine | |
| CN110706225B (en) | Tumor identification system based on artificial intelligence | |
| Hou | Breast cancer pathological image classification based on deep learning | |
| Li et al. | Transfer learning based classification of cervical cancer immunohistochemistry images | |
| CN116188423A (en) | Super-pixel sparse and unmixed detection method based on pathological section hyperspectral image | |
| Kaur et al. | Optimized multi threshold brain tumor image segmentation using two dimensional minimum cross entropy based on co-occurrence matrix | |
| Nateghi et al. | Automatic detection of mitosis cell in breast cancer histopathology images using genetic algorithm | |
| Rahman et al. | BreastMultiNet: A multi-scale feature fusion method using deep neural network to detect breast cancer | |
| Jamaluddin et al. | Tumor detection and whole slide classification of H&E lymph node images using convolutional neural network | |
| Mustafa et al. | Comparison of detection method on malaria cell images | |
| Vidyarthi et al. | Classification of breast microscopic imaging using hybrid CLAHE-CNN deep architecture | |
| Kriti et al. | A review of segmentation algorithms applied to B-mode breast ultrasound images: A characterization approach | |
| Manivasagan et al. | A survey on various kinds of image segmentation of cancer images using fuzzy image processing methods | |
| Bao et al. | Computational and mathematical methods in medicine | |
| Jiang et al. | Parallel ‘same’and ‘valid’convolutional block and input-collaboration strategy for histopathological image classification | |
| Gupta et al. | Breast cancer identification using feature level fusion and hybrid GA-PSO optimized 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |