CN111612735A - Lung nodule image classification method based on information fusion safety semi-supervised clustering - Google Patents
Lung nodule image classification method based on information fusion safety semi-supervised clustering Download PDFInfo
- Publication number
- CN111612735A CN111612735A CN202010269007.9A CN202010269007A CN111612735A CN 111612735 A CN111612735 A CN 111612735A CN 202010269007 A CN202010269007 A CN 202010269007A CN 111612735 A CN111612735 A CN 111612735A
- Authority
- CN
- China
- Prior art keywords
- clustering
- sample
- membership
- samples
- marked
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000004927 fusion Effects 0.000 title claims abstract description 32
- 206010056342 Pulmonary mass Diseases 0.000 title claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 19
- 230000002685 pulmonary effect Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000005192 partition Methods 0.000 claims description 9
- 230000010354 integration Effects 0.000 claims description 5
- 239000008186 active pharmaceutical agent Substances 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 3
- 239000000956 alloy Substances 0.000 claims description 2
- 229910045601 alloy Inorganic materials 0.000 claims description 2
- 230000015572 biosynthetic process Effects 0.000 claims description 2
- 210000004072 lung Anatomy 0.000 claims description 2
- 239000003550 marker Substances 0.000 claims description 2
- 230000000877 morphologic effect Effects 0.000 claims description 2
- 238000003786 synthesis reaction Methods 0.000 claims description 2
- 238000013210 evaluation model Methods 0.000 abstract description 2
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 6
- 201000005202 lung cancer Diseases 0.000 description 6
- 208000020816 lung neoplasm Diseases 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 206010028980 Neoplasm Diseases 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000004195 computer-aided diagnosis Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 201000011510 cancer Diseases 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000034994 death Effects 0.000 description 1
- 231100000517 death Toxicity 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 230000003211 malignant effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/257—Belief theory, e.g. Dempster-Shafer
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
- G06T2207/30064—Lung nodule
Abstract
The invention discloses a lung nodule image classification method based on information fusion safety semi-supervised clustering. Then, a risk degree evaluation model of the marked sample is established. And constructing and selecting a plurality of base clustering methods by utilizing the idea of ensemble learning, fusing the division results of the base clustering methods on the marked samples by a D-S evidence theory method, and obtaining the risk degree of the marked samples according to the fusion results. Next, an optimization model based on graph regularization is constructed. And finally, solving the optimization model by adopting an iterative optimization strategy to obtain a clustering result. The method solves the problem of safe use of the labeled sample, and improves the accuracy and robustness of classification of the pulmonary nodules.
Description
Technical Field
The invention relates to a semi-supervised clustering method for robustness of a labeled sample, in particular to a robust semi-supervised clustering algorithm based on DS evidence theory fusion weighting, and belongs to the field of data mining of medical images.
Background
The world health organization showed in 2018 the latest cancer report worldwide: the incidence rate of lung cancer is highest in the whole world (210 ten thousand new patients account for 11.6 percent of the incidence rate of various cancers) and the mortality rate (180 ten thousand deaths account for 18.4 percent of the mortality rate of various cancers). The early manifestation of lung cancer is almost without any symptoms, resulting in the patients being found most of the time in an advanced stage, and therefore, it is important to find lung cancer early to improve survival rate. Lung cancer generally exists in the form of nodules in early stage, and the lung nodules are found immediately to greatly improve the cure rate of lung cancer patients. For doctors, because of subjectivity and other factors, the selection of lung nodules from a large number of CT images is very likely to cause misdiagnosis and missed diagnosis, so computer-aided diagnosis (CAD) is very important. Semi-supervised learning is a hot point of research in recent years, and a semi-supervised FCM algorithm is taken as one of typical algorithms in a semi-supervised clustering algorithm, is favored by broad scholars due to lower complexity and better application effect in practical problems.
The benign and malignant classification of lung nodules is of great significance for early detection and diagnosis of lung cancer. In practical applications, the number of images of the marked sample is small, and its label is not always accurate. In the semi-supervised clustering process, the labeled samples guide clustering, and the accuracy of the labeled samples is extremely high. Gan et al have verified that tag information can lead to a degradation in semi-supervised learning performance. How to correctly utilize a small number of labeled samples which may have wrong labels makes the performance of the semi-supervised clustering method more robust is worthy of research and exploration. Therefore, the invention tries to design a semi-supervised clustering method which is robust to the marked samples, and under the condition of wrong labels, the clustering performance is more excellent and more stable compared with the original unsupervised clustering and semi-supervised clustering.
Disclosure of Invention
The invention provides a lung nodule image classification method based on information fusion safety semi-supervised clustering, aiming at the defect that the final classification effect is possibly reduced because the risk of marking samples is not considered in the traditional lung nodule classification method based on semi-supervised clustering.
Firstly, a risk degree evaluation model of a marked sample is established. And constructing a plurality of base clustering algorithms by utilizing the idea of ensemble learning, obtaining the clustering result of each base clustering algorithm on the data set, obtaining the effectiveness of each base clustering algorithm through a clustering effectiveness index, and selecting the base clustering algorithm with good clustering effect. And fusing the clustering information of each base clustering algorithm on the marked sample through a D-S evidence theory, and obtaining the risk degree of the marked sample according to the result of information fusion. Then, a regularization-based optimization model is constructed. And constructing a regular term based on the relation matrix and the k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain an optimization model. And finally, solving the optimization model by adopting an iterative optimization algorithm to obtain a clustering result.
The invention relates to a pulmonary nodule method based on information fusion safety semi-supervised clustering, which comprises the following steps:
the method comprises the following steps: inputting labeled and unlabeled lung nodule images;
step two: preprocessing a lung nodule image and extracting characteristics;
step three: constructing a basis clustering algorithm, and pre-clustering the data set;
step four: estimating the reliability of the basis clustering algorithm by adopting an internal validity index according to a pre-clustering result, and selecting W basis clusters as information fusion evidences;
step five: calculating an evidence fusion partition matrix of the marked samples by using a fuzzy clustering integration method based on a D-S evidence theory;
step six: calculating the risk degree of the marked sample according to the fusion result, and converting the risk degree into the safety degree by utilizing a sine function;
step seven: constructing a local k-neighbor graph with the aim of limiting the labeled sample output with low confidence to the output of adjacent unlabeled samples;
step eight: constructing a regular term based on the relation matrix and the k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain an optimization model;
step nine: solving the optimization problem by adopting an iterative optimization method;
step ten: and judging the category of the unlabeled sample to realize the classification of the lung nodule image.
Compared with the traditional semi-supervised clustering method, the method has the advantages that the D-S evidence theory is utilized to fuse a plurality of clustering algorithm results to obtain the labeled sample risk degree, and the labeled samples with low confidence coefficient are limited to be the output of the adjacent samples by constructing the local graph, so that each labeled sample can be safely and reasonably used, and the clustering is more accurate and robust.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention.
Detailed Description
The present invention is further illustrated with reference to the accompanying figure 1, it being understood that these examples are intended to be illustrative only and not to be limiting as to the scope of the invention, which is to be given the full breadth of the appended claims and any and all equivalent modifications thereof which would occur to persons skilled in the art upon reading the present specification.
To better illustrate the objects and advantages of the present invention, embodiments of the method of the present invention are described in further detail below with reference to the accompanying drawings and examples.
The method comprises the following steps: inputting feature data sets of labeled and unlabeled lung nodule images;
subset of labeled samples of the input data set: xl=[x1...xl]The corresponding label is yk∈ { 1...., c }, unlabeled sample subset Xu=[xl+1...xn];
Step two: preprocessing a lung nodule image and extracting characteristics;
and (3) eliminating the noise of the CT image by adopting a self-adaptive filter, segmenting pulmonary nodules of the lung CT image by using a rapid FCM clustering method, and extracting gray scale features, texture features, morphological features and boundary features.
Step three: constructing a basis clustering algorithm, and pre-clustering the data set;
adopting a homomorphic clustering integration method; taking the fuzzy C mean value as a unique clustering device based on different fuzziness and different kernelsForming a plurality of base clusters, and recording the clustering result as pi ═ pi1,Π2,...,ΠW′II, whereinwRepresenting the w-th cluster member in the cluster set pi; the clustering result of each base cluster contains the clustering information of the samples, and the clustering information of the marked samples is the membership degree of the marked samples;
step four: estimating the reliability of the basis clustering algorithm by adopting an internal validity index according to a pre-clustering result, and selecting W basis clusters as information fusion evidences;
clustering has randomness, and when information is processed, the information needs to be distinguished according to the clustering effectiveness of each base clustering algorithm. If not differentiated, accepting the utilization in the same way can introduce errors into the results. Therefore, it is necessary to select an algorithm with a good clustering effect according to a certain method. Therefore, the influence of conflict information can be reduced while more useful information is kept, and the fusion result is more accurate. The quality of the clustering effect can not be directly judged, and the accuracy of clustering is indirectly judged by adopting a contour coefficient method. And selecting W basic clusters with high clustering accuracy as information fusion evidence.
Step five: calculating to obtain an evidence fusion partition matrix of the marked samples by using a fuzzy clustering integration method based on a DS evidence theory;
DS evidence theory generally includes a recognition framework, a basic confidence assignment (BPA) function, and an evidence synthesis 3 part.
Assume dataset label yk∈ { 1...., c }, it is easy to know that the mark sample identification frame is { class 1...., class c }. for mark sample xkThe BPA function value is the sample membership degree of a base clustering algorithm (namely a fuzzy C-means clustering algorithm) to the data set, and the sample membership degree of each base clustering is fused through a D-S evidence theory to obtain an evidence fusion partition matrix for marking samplesWhereinEvidence fusionDegree of membership;
step six: obtaining the risk degree of the marked sample according to the fusion result, and converting the risk degree into the safety degree by utilizing a sine function;
in the process of measuring the risk of the prior information, the (c +1) th row vector of the evidence fusion partition matrix is not considered. Marking the risk degree of the marked sample as P ═ Pk]l。F=[fik]c×lIndicating the ambiguity of the marked sample when sample xkFor marking samples and belonging to class i fik1, otherwise equal to 0.
To formulateikFusion membership of 1 hour with evidenceIs used as the marker sample xkFirst risk value ofThe calculation formula is as follows:
when f isikWhen the number is equal to 1, the alloy is put into a container,
to further pull apart the risk between the marked samples, a second risk value is definedAnd defuzzifying the evidence fusion partition matrix according to the maximum membership principle, wherein if the result is consistent with the given label of the marked sample, the sample is relatively safe, otherwise, the sample is unsafe. For the relatively safe marked sample, the second risk value is the evidence fusion membership valueDegree of sub-maximum membershipThe difference of (a). For unsafe marked samples, the second windThe risk value is the maximum membership valueAndthe difference of (a). The calculation formula is as follows:
marking sample Risk pkThe calculation formula of (a) is as follows:
converting the marked sample risk degree into a safety degree, wherein the calculation formula is as follows:
step seven: constructing a local k-neighbor graph with the aim of limiting labeled sample outputs with low confidence to those of neighboring samples;
constructing a local neighborhood graph of the marked sample, and then weighting W ═ W of the local graph edgekr]n×nCan be calculated as:
wherein N isp(xk) Finger xkP data of nearest neighbor, xkTo mark sample points, xrσ represents the width parameter of the gaussian kernel for the neighboring sample points.
Step eight: and constructing a regular term based on the relation matrix and the local k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain an optimization model.
Constructing a regular term based on a relation matrix and a local k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain a safe semi-supervised fuzzy C-means clustering target function based on information fusion as follows:
the limiting conditions are as follows:μirrepresenting membership of unlabeled samples, and m represents ambiguity;
Step nine: solving the optimization problem by adopting an iterative optimization method;
by minimizing the above optimization problem, an optimal solution can be obtained. To simplify the calculation, the value of m is set to 2. The method solves the sample membership degree and the clustering center by adopting a Lagrange multiplier method.
Membership of labeled samples:
membership mu of unlabeled samplesir:
cluster center vi:
And obtaining a final membership matrix U and a clustering center V through iterative calculation. When in useOr a maximum number of iterations is reached, where t is the current number of iterations, η is a set threshold,and (4) representing the clustering result of the t-th time.
Step ten: and judging the category of the unlabeled sample, and realizing the classification of the lung nodule picture.
And after obtaining the membership matrix U, defuzzifying according to the maximum membership principle to obtain the category of the unlabeled sample, and finally classifying the pictures to obtain a result.
Claims (3)
1. The lung nodule image classification method based on information fusion safe semi-supervised clustering is characterized by comprising the following steps of:
the method comprises the following steps: inputting labeled and unlabeled lung nodule images;
subset of labeled samples of the input data set: xl=[x1...xl]The corresponding label is yk∈ { 1...., c }, unlabeled sample subset Xu=[xl+1...xn];
Step two: preprocessing a lung nodule image and extracting characteristics;
step three: constructing a basis clustering algorithm, and pre-clustering the data set;
adopting a homomorphic clustering integration method; will blur CThe mean value is used as a unique clustering device, a plurality of base clusters are formed based on different fuzziness degrees and different kernels, and a clustering result is recorded as pi ═ pi1,Π2,...,ΠW′II, whereinwRepresenting the w-th cluster member in the cluster set pi; the clustering result of each base cluster contains the clustering information of the samples, and the clustering information of the marked samples is the membership degree of the marked samples;
step four: estimating the reliability of the basis clustering algorithm by adopting an internal validity index according to a pre-clustering result, and selecting W basis clusters as information fusion evidences;
step five: calculating an evidence fusion partition matrix of the marked samples by using a fuzzy clustering integration method based on a D-S evidence theory;
the DS evidence theory comprises three parts, namely a recognition framework, a basic confidence degree assignment function and evidence synthesis;
assume dataset label yk∈ {1, 1.. multidot.c }, then marking the sample identification frame as { class 1, 1.. multidot.c }, and for the marked sample xkThe basic confidence degree assignment function value is the sample membership degree of the base clustering algorithm to the data set, and the sample membership degree of each base clustering is fused through the D-S evidence theory to obtain an evidence fusion partition matrix for marking samplesWhereinFusing membership degrees of the evidences;
step six: calculating the risk degree of the marked sample according to the fusion result, and converting the risk degree into the safety degree by utilizing a sine function;
in the process of measuring the risk of the prior information, the (c +1) th row vector of the evidence fusion partition matrix is not considered; marking the risk degree of the marked sample as P ═ Pk]l;F=[fik]c×lIndicating the ambiguity of the marked sample when sample xkFor marking samples and belonging to class i fik1, otherwise equal to 0;
preparingWill f isikFusion membership of 1 hour with evidenceIs used as the marker sample xkFirst risk value ofThe calculation formula is as follows:
when f isikWhen the number is equal to 1, the alloy is put into a container,
to further pull apart the risk between the marked samples, a second risk value is definedAccording to the maximum membership principle, defuzzifying the evidence fusion partition matrix, if the result is consistent with the given label of the marked sample, indicating that the sample is relatively safe, otherwise, indicating that the sample is unsafe; for the relatively safe marked sample, the second risk value is the evidence fusion membership valueDegree of sub-maximum membershipA difference of (d); for unsafe marked samples, the second risk value is the maximum membership valueAnda difference of (d); the calculation formula is as follows:
marking sample Risk pkThe calculation formula of (a) is as follows:
converting the marked sample risk degree into a safety degree, wherein the calculation formula is as follows:
step seven: constructing a local k-neighbor graph with the aim of limiting the labeled sample output with low confidence to the output of adjacent unlabeled samples;
constructing a local neighborhood graph of the marked sample, and then weighting W ═ W of the local graph edgekr]n×nThe calculation is as follows:
wherein N isp(xk) Finger xkP data of nearest neighbor, xkTo mark sample points, xrIs a neighboring sample point, and sigma represents a width parameter of the Gaussian kernel function;
step eight: constructing a regular term based on the relation matrix and the k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain an optimization model;
constructing a regular term based on a relation matrix and a local k-nearest neighbor graph, and embedding the regular term into the existing clustering algorithm to obtain a safe semi-supervised fuzzy C-means clustering target function based on information fusion as follows:
the limiting conditions are as follows:μirrepresenting membership of unlabeled samples, and m represents ambiguity;
wherein λ is1,λ2For regularization parameter, U ═ μik]c×nAs a membership matrix, distance metricviRepresenting a cluster center;
step nine: solving the optimization problem by adopting an iterative optimization method;
step ten: judging the category of the unlabeled sample to realize classification of the lung nodule image;
and after obtaining the membership matrix U, defuzzifying according to the maximum membership principle to obtain the category of the unlabeled sample, and finally classifying the pictures to obtain a result.
2. The lung nodule image classification method based on information fusion safety semi-supervised clustering as claimed in claim 1, wherein: preprocessing the lung nodule image and extracting characteristics in the second step, which specifically comprises the following steps: and (3) eliminating the noise of the CT image by adopting a self-adaptive filter, segmenting pulmonary nodules of the lung CT image by using a rapid FCM clustering method, and extracting gray scale features, texture features, morphological features and boundary features.
3. The lung nodule image classification method based on information fusion safety semi-supervised clustering as claimed in claim 1, wherein: solving the optimization problem by adopting an iterative optimization method; the method specifically comprises the following steps:
solving sample membership and a clustering center by adopting a Lagrange multiplier method;
membership of labeled samples:
membership mu of unlabeled samplesir:
cluster center vi:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010269007.9A CN111612735A (en) | 2020-04-08 | 2020-04-08 | Lung nodule image classification method based on information fusion safety semi-supervised clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010269007.9A CN111612735A (en) | 2020-04-08 | 2020-04-08 | Lung nodule image classification method based on information fusion safety semi-supervised clustering |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111612735A true CN111612735A (en) | 2020-09-01 |
Family
ID=72202151
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010269007.9A Pending CN111612735A (en) | 2020-04-08 | 2020-04-08 | Lung nodule image classification method based on information fusion safety semi-supervised clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111612735A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103700108A (en) * | 2013-12-24 | 2014-04-02 | 西安电子科技大学 | Image segmentation method adopting semi-supervised RFLICM (Robust Fuzzy Local Information C-Means) clustering on basis of seed set |
CN108038511A (en) * | 2017-12-25 | 2018-05-15 | 江苏江大智慧科技有限公司 | Cluster hypothesis is corrected to be unified into constraining semisupervised classification method |
CN110263804A (en) * | 2019-05-06 | 2019-09-20 | 杭州电子科技大学 | A kind of medical image dividing method based on safe semi-supervised clustering |
-
2020
- 2020-04-08 CN CN202010269007.9A patent/CN111612735A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103700108A (en) * | 2013-12-24 | 2014-04-02 | 西安电子科技大学 | Image segmentation method adopting semi-supervised RFLICM (Robust Fuzzy Local Information C-Means) clustering on basis of seed set |
CN108038511A (en) * | 2017-12-25 | 2018-05-15 | 江苏江大智慧科技有限公司 | Cluster hypothesis is corrected to be unified into constraining semisupervised classification method |
CN110263804A (en) * | 2019-05-06 | 2019-09-20 | 杭州电子科技大学 | A kind of medical image dividing method based on safe semi-supervised clustering |
Non-Patent Citations (2)
Title |
---|
依玉峰;高立群;郭丽;: "基于相似类合并FCM在图像分割中的应用" * |
姜婷;袭肖明;岳厚光;: "基于分布先验的半监督FCM的肺结节分类" * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Cao et al. | A multi-kernel based framework for heterogeneous feature selection and over-sampling for computer-aided detection of pulmonary nodules | |
Wang et al. | Weakly supervised learning for whole slide lung cancer image classification | |
Shao et al. | A detection approach for solitary pulmonary nodules based on CT images | |
Yuan et al. | Discriminative joint-feature topic model with dual constraints for WCE classification | |
Chen et al. | Pulmonary nodule detection on chest radiographs using balanced convolutional neural network and classic candidate detection | |
CN116030325A (en) | Lung nodule CT image recognition method based on deep hybrid learning framework | |
CN111968124A (en) | Shoulder musculoskeletal ultrasonic structure segmentation method based on semi-supervised semantic segmentation | |
Cao et al. | A ℓ2, 1 norm regularized multi-kernel learning for false positive reduction in lung nodule CAD | |
Basha et al. | An effective and robust cancer detection in the lungs with BPNN and watershed segmentation | |
CN110263804B (en) | Medical image segmentation method based on safe semi-supervised clustering | |
Tejero-de-Pablos et al. | Texture-based classification of significant stenosis in CCTA multi-view images of coronary arteries | |
Xu et al. | A new object detection algorithm based on yolov3 for lung nodules | |
Kumar et al. | A methodical exploration of imaging modalities from dataset to detection through machine learning paradigms in prominent lung disease diagnosis: a review | |
Fazilov et al. | Patch-based lesion detection using deep learning method on small mammography dataset | |
Naik et al. | Prediction of malignancy in lung nodules using combination of deep, fractal, and Gray-level Co-occurrence matrix features | |
Wang | Automatic mushroom species classification model for foodborne disease prevention based on vision transformer | |
Bhatt et al. | Image retrieval using bag-of-features for lung cancer classification | |
Antonelli et al. | Computer-aided detection of lung nodules based on decision fusion techniques | |
Song et al. | Weakly-supervised classification of pulmonary nodules based on shape characters | |
Xu et al. | Identification of benign and malignant lung nodules in CT images based on ensemble learning method | |
Pan et al. | A review of machine learning approaches, challenges and prospects for computational tumor pathology | |
CN111612735A (en) | Lung nodule image classification method based on information fusion safety semi-supervised clustering | |
CN111415350B (en) | Colposcope image identification method for detecting cervical lesions | |
Liu et al. | SASG-GCN: self-attention similarity guided graph convolutional network for multi-type lower-grade glioma classification | |
Zanwar et al. | A comprehensive survey on soft computing based optical character recognition techniques |
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 | ||
AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20240507 |