CN111904417B - Ultra-wideband microwave early breast cancer detection device based on support vector machine - Google Patents

Ultra-wideband microwave early breast cancer detection device based on support vector machine Download PDF

Info

Publication number
CN111904417B
CN111904417B CN202010641919.4A CN202010641919A CN111904417B CN 111904417 B CN111904417 B CN 111904417B CN 202010641919 A CN202010641919 A CN 202010641919A CN 111904417 B CN111904417 B CN 111904417B
Authority
CN
China
Prior art keywords
module
data set
support vector
characteristic
vector machine
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.)
Active
Application number
CN202010641919.4A
Other languages
Chinese (zh)
Other versions
CN111904417A (en
Inventor
肖夏
刘冠聪
宋航
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN202010641919.4A priority Critical patent/CN111904417B/en
Publication of CN111904417A publication Critical patent/CN111904417A/en
Application granted granted Critical
Publication of CN111904417B publication Critical patent/CN111904417B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves 
    • A61B5/0507—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves  using microwaves or terahertz waves
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/43—Detecting, measuring or recording for evaluating the reproductive systems
    • A61B5/4306—Detecting, measuring or recording for evaluating the reproductive systems for evaluating the female reproductive systems, e.g. gynaecological evaluations
    • A61B5/4312—Breast evaluation or disorder diagnosis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • 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
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Evolutionary Computation (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Medical Informatics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Psychiatry (AREA)
  • Physiology (AREA)
  • Fuzzy Systems (AREA)
  • Signal Processing (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Gynecology & Obstetrics (AREA)
  • Reproductive Health (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

The invention relates to an ultra-wideband microwave early breast cancer detection device based on a support vector machine, which comprises a feature extraction module, a data set construction module, a feature selection module and a support vector machine learning classification module, wherein the feature extraction module is used for acquiring an original signal from a breast to be detected and acquiring corresponding specific features; constructing an original characteristic data set through a data set construction module; optimizing the original characteristic data set through a characteristic selection module; and finally, classifying the optimized data set by using a support vector machine module to form an effective detection model. The characteristic extraction module acquires signal characteristics by combining a modal decomposition method with statistical characteristics; and the characteristic selection module reduces the data dimension by projecting the characteristic data set to a subspace capable of keeping the maximum variance of the signal set by adopting a principal component analysis method, and removes redundant characteristics. And a Gaussian SVM method is used in a support vector machine learning classification module.

Description

Ultra-wideband microwave early breast cancer detection device based on support vector machine
Technical Field
The invention belongs to the field of biomedical detection and ultra wide band microwave detection, and relates to a novel ultra wide band microwave early breast cancer nondestructive detection device.
Background
Because the traditional breast cancer detection methods have certain disadvantages and shortcomings, more and more researchers and research institutions develop a lot of related research works on developing a novel early-stage breast cancer screening and detecting method in recent years. Because the dielectric properties of benign breast tissue and malignant breast tissue have obvious difference in the microwave frequency range, and the ultra-wideband antenna often has lower radiation level and cannot damage human bodies, the application of the ultra-wideband technology as a nondestructive detection technology to the detection and screening of breast cancer is more and more possible. Currently, some breast tumor detection methods and detection devices based on ultra-wideband technology exist in the world, but most of the breast tumor detection methods and detection devices realize corresponding functions based on simulation models or simple solid models. At present, some problems still exist in the complex application and clinical application of the ultra-wideband microwave breast tumor, and a large amount of related researches are urgently needed to be carried out and verified. Therefore, the method for realizing the nondestructive detection of the breast tumor through the ultra-wideband microwave has obvious advantages and great potential, and the research in the field can be divided into two aspects of an imaging algorithm and a detection algorithm at present. The two aspects have the same signal acquisition principle, and different functions and targets are realized according to different processing methods after the signals are transmitted and received by the antennas for recording and detecting the whole breast. The imaging algorithm inverts the propagation path of the signal to finally construct an internal image of the breast, and the internal image can be accurately shown in an imaging image when a tumor is contained in the detected object. The detection algorithm extracts a feature data set formed by specific features from the detected object acquisition signals in a learning and classifying mode, and then accurately judges whether the tumor exists in the detected object.
The invention belongs to one of detection devices, and aims at the important requirement of accurately judging whether tumors exist in breasts, relevant contents are researched, and an implementation scheme of an algorithm is analyzed, so that the early breast cancer diagnosis device based on a support vector machine is invented.
Disclosure of Invention
The invention provides a device capable of accurately detecting early breast tumors. The technical scheme of the invention is as follows:
an ultra-wideband microwave early breast cancer detection device based on a support vector machine comprises a feature extraction module, a data set construction module, a feature selection module and a support vector machine learning classification module, wherein an original signal is obtained from a breast to be detected through the feature extraction module, and corresponding specific features are obtained; constructing an original characteristic data set through a data set construction module; optimizing the original characteristic data set through a characteristic selection module; finally, a support vector machine module is used for classifying the optimized data set to form an effective detection model, and the method is characterized in that a feature extraction module acquires signal features by combining a modal decomposition method with statistical characteristics; and the characteristic selection module reduces the data dimension by projecting the characteristic data set to a subspace capable of keeping the maximum variance of the signal set by adopting a principal component analysis method, and removes redundant characteristics. And a Gaussian SVM method is used in a support vector machine learning classification module.
Preferably, the detection steps implemented according to the echo signals obtained by the antenna array detecting the breast are as follows:
(1) carrying out normalization processing on echo signals of each antenna by using a minimum and maximum normalization algorithm;
(2) decomposing and analyzing a total M groups of echo signals acquired from the echo signals detected from a plurality of breasts by using a feature extraction module formed by integrating methods of empirical mode decomposition, effective feature selection, statistical characteristic extraction and the like, and extracting all specific features of tumor signals and tumor-free signals; the method comprises the following steps:
(3) all the specific characteristics form an original characteristic data set through a data set construction module, the characteristics of the tumor echo signals are set as a positive sample, and a data label is set as 1; the characteristics of the tumor-free echo signals are set as negative samples, and the data label is set as-1;
(4) normalizing the data in the original characteristic data set again by using a minimum and maximum normalization algorithm to eliminate dimensional influences of different characteristics;
(5) in the feature selection module, a principal component analysis method is used for feature selection, M-dimensional data is changed to N-dimensional data through a corresponding algorithm, M is larger than N, and finally a simplified optimal data set with redundant variables removed is obtained;
(6) and (3) building and training a detection model on the optimized data set by using a support vector machine learning classification module based on a Gaussian kernel function.
The feature extraction step of the feature extraction module comprises the following steps:
1) each 1 group of echo signals are decomposed into a plurality of groups of signal components through integrated empirical mode decomposition, and the component containing the most sufficient original signal information is judged and reserved through a correlation coefficient at the second stage;
2) taking 0.03 as a threshold value selected by the effective component, judging the IMF component with the correlation coefficient larger than the threshold value as the effective signal component, and discarding the rest signal components;
3) in the statistical characteristic extraction, the characteristic extraction based on 6 statistical characteristics of mean, variance, standard deviation, maximum value, energy entropy and information entropy is carried out on effective signal components, and specific characteristics are obtained from each 1 group of single-channel echo signals.
Drawings
FIG. 1 is a schematic diagram of an antenna array arrangement
FIG. 2 is a diagram of a single channel signal decomposition result
FIG. 3 work flow diagram
Detailed Description
The invention aims to overcome the defects of the existing detection technology and provides a device for detecting whether a patient suffers from early breast cancer. The device is an electronic device, comprising a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the method steps realized when the processor executes the program are as follows: a large number of features are obtained from echo signals based on a feature extraction module, and an original feature data set containing all tumor tissue specific features and normal tissue specific features is constructed by utilizing a data set construction module. And finally, generalizing and classifying the tumor-containing features and the tumor-free features which contain labels in the feature data set by using a support vector machine module to obtain a corresponding classification model. And a detection result is obtained for the echo signal to be detected based on the classification model, so that the early breast cancer can be accurately detected.
The technical scheme of the invention is specifically explained by the following practical working flow:
on the basis of an original characteristic data set which contains a large number of breast tumor tissue and normal tissue characteristics and is obtained by a signal decomposition and selection and specific characteristic extraction and data set construction module of a characteristic extraction module, a PCA method of the characteristic selection module is used for calculating and analyzing the characteristic data set to obtain an optimized data set which is lower in dimensionality and more representative in data. And further realizing a detection model of the tumor condition and the tumor-free condition by using an SVM module based on a Gaussian kernel function on the basis of optimizing the data set, and distinguishing and detecting the echo signal to be detected on the basis of the detection model. The method comprises the following specific steps:
1. an antenna array arrangement for detecting the breast is shown in fig. 1, in which case an antenna array of 8 antennas, a1-A8, is used to cover substantially the entire breast area, enabling efficient detection of the breast to be detected. Because the work flow of the antenna array is that 1 antenna sends 8 antennas for receiving, after a complete detection flow, the square group echo signals with the number of the antennas are obtained;
2. echo signals in the case of self-emission and self-reception are removed from the collected sets of echo signals, and the echo signals used in the feature extraction module in the case of fig. 1 are (8 × 8-8) ═ 56 sets. The tumor-present and tumor-free signals obtained from several breasts eventually together constitute M sets of echo signals, which are adjusted to a ratio of 1: 1. Changing M groups of echo signals into an interval [0,1] in a maximum and minimum normalization mode;
Figure BDA0002571459920000031
in the formula, xc(t) is the normalized time series, x (t) is the original signal time series, xmaxTo normalize the maximum value of the time series beforeminIs the minimum value of the time series before normalization;
3. and (3) performing feature extraction on the M groups of echo signals by using a feature extraction module, wherein the feature extraction module comprises three stages of integrated empirical mode decomposition, effective component selection and statistical characteristic calculation.
Each 1 group of echo signals are decomposed into a plurality of groups of signal components in the first stage by integrating empirical mode decomposition, and the echo signals are decomposed by setting appropriate decomposition parameters in consideration of calculation cost according to use experience and recommendation in some documents. The method can perform self-adaptive decomposition on any nonlinear and non-stationary signals to obtain a plurality of Intrinsic Mode Functions (IMFs). Decomposition principle of EEMD andthe formula is as follows, wherein rj(t) is an IMFS component.
Figure BDA0002571459920000032
The noise standard deviation is selected to be 0.2 times of the standard deviation of the original signal, and the integration times is set to be 100 times as the decomposition parameters of the EEMD to decompose all echo signals. One of the single-channel decomposition results is shown in fig. 2. A total of 12 IMF signal components C1-C12 and 1 residual term r12 were generated.
Secondly, in order to extract effective characteristics from a plurality of signal components, the influence of characteristics of irrelevant signal components is reduced. And a pre-selection step of effective signal components is adopted before the extraction of the specific features, and is realized by calculating the Pearson correlation coefficient of each signal component and the original signal. Let x (k) and rj(k) Two sequences of length n, whose Pearson correlation coefficient is defined as follows, where rj(k) Corresponding to the jth IMF component. Through a large number of decomposition experiments and attempts, the threshold value selected by the invention by taking 0.03 as the effective component is adopted. I.e. the retained correlation coefficient ccjIMF components greater than 0.03, and the 4 signal components in fig. 2 having correlation values greater than the threshold are retained, and the remaining signal components are discarded.
Figure BDA0002571459920000033
Wherein x (k) and rj(k) Respectively corresponding to two sequences with length of m
Figure BDA0002571459920000034
And
Figure BDA0002571459920000035
corresponding to the average of the two sequences, respectively, and j corresponds to the jth IMF component.
Calculating the correlation coefficient of each signal component and the original signal through the correlation coefficient is shown in table 1, and the effective signal components are screened out by combining the set corresponding threshold values, and 4 most representative signal components C5-C8 are selected from the decomposition result. After statistical feature extraction, each signal component will get 6 features. Finally, each single-channel echo signal can generate 24 features by the feature extraction method provided by the invention.
TABLE 1 correlation coefficient of each signal component with a single-channel original signal
Figure BDA0002571459920000036
Figure BDA0002571459920000041
In the third-stage statistical characteristic extraction, feature extraction based on 6 statistical characteristics of mean, variance, standard deviation, maximum value, energy entropy and information entropy is carried out on each signal component obtained in the second stage, and finally 24 specific features are obtained from each 1 group of single-channel echo signals;
for IMF components r of length nj＝{rj(1),rj(2),rj(3),…,rj(n), the mean, variance, standard deviation, maximum value, energy entropy and information entropy can be expressed by the following mathematical formulas. The characteristics corresponding to a plurality of groups of back scattering signals generated by the antenna array are obtained by a characteristic statistical characteristic extraction method, and an original characteristic data set consisting of characteristics of early breast tumor tissues and normal breast tissues can be constructed.
Mean value (mu)j)
Figure BDA0002571459920000042
② variance (D)j)
Figure BDA0002571459920000043
③ standard deviation (delta)j)
Dj＝δj 2
Maximum value (Max)j)
Maxj＝Max{rj(1),rj(2),rj(3),…,rj(n)}
Energy Entropy (EE)j)
Figure BDA0002571459920000044
Information Entropy (IE)j)
Figure BDA0002571459920000045
1. And splicing 24 features of all M groups of echo signals by using a data set construction module to form an original feature data set with the scale of M multiplied by 24. Where M is a row vector corresponding to each 1 group of echo signals and 24 is a column vector corresponding to 1 specific feature in each 1 group of echo signals. Meanwhile, setting the label with tumor signal characteristics as +1 and the label without tumor signal characteristics as-1;
2. and (3) because the data units in the training set are different and the numerical magnitudes are also different, the maximum and minimum normalization method which is the same as the step (2) is adopted again to perform dimension elimination processing on the training set, and the quality of the training set is further improved. All 24 features extracted from each 1 group of echo signals in the original feature data set are normalized;
3. the reduced data set is obtained using a feature selection module consisting of a PCA feature selection method. For X ∈ RM×24,Xt＝{xt(1),xt(2),xt(3),…,xt(24) First calculate the mean value of each column vector:
Figure BDA0002571459920000046
the covariance matrix C of the original dataset X is calculated by means of the mean:
Figure BDA0002571459920000051
calculating an eigenvalue λ of the covariance matrix CiAnd a feature vector vi(i ═ 1,2, …,24), wherein eigenvalues are arranged from large to small, and the first p principal components are selected according to a certain criterion to respectively construct a diagonal matrix Λ and an eigenvector matrix V:
Cvi＝λivi
Figure BDA0002571459920000052
finally, the original feature data set X is transformed into an optimized data set P with lower dimensionality and more representative by using a new feature vector matrix V, and the optimized data set can approximately represent most characteristics of the original feature data set, but has stronger feature characterization capability and smaller data set dimensionality:
P＝VTX
4. and a training model generated by a support vector machine learning module of the Gaussian SVM based on the optimized data set is used as a detection scheme, and the new sample to be detected after the same-proportion normalization is applied to the classification model. If the label of the sample data to be detected obtained by calculating the classification model of the support vector machine is +1, indicating that the early breast tumor exists in the detected breast; if the label of the sample data to be detected calculated by the classification model of the support vector machine is-1, the detected breast does not contain early breast tumor and is in a healthy state. Compared with a neural network, the support vector machine is not easy to have the over-fitting problem and is not influenced by the space dimension of a data set, the input vector is projected to a high-dimensional characteristic space through nonlinear mapping, a nonlinear decision boundary in an original space is established in a linear model of a new space, and a detection model with better performance is further established.
The workflow of the present invention is summarized as follows:
the workflow block diagram of the present invention is shown in fig. 3.
(1) Obtaining 4 effective signal components from a plurality of groups of signals to be detected through an integrated empirical mode decomposition and effective component selection stage in a feature extraction module, and extracting 24 specific features from each 1 group of echo signals through 6 statistical characteristic feature extraction methods;
(2) constructing all the specific characteristics into an Mx 24 original characteristic data set through a data set construction module, and adding category labels for characteristics with tumors and characteristics without tumors respectively;
(3) and optimizing and selecting the M multiplied by 24 original data set by using a characteristic selection module and adopting a proper main component number (the dimensionality of the data set after dimensionality reduction). Obtaining an optimized dataset of M x (24-K);
(4) learning and classifying the optimized data set through a Gaussian SVM learning classification module to obtain a corresponding detection classification model;
(5) and acquiring echo signals from the new breast to be detected, and applying the echo signals to an early breast cancer detection classification model after feature extraction to obtain a detection result.
The method is an early breast cancer intelligent detection device based on a support vector machine, and can accurately analyze and judge signals obtained by a breast to be detected under the conditions of a large number of specific characteristics under the condition of tumor and specificity under the condition of no tumor to obtain an effective early breast tumor detection model. Is a simple, stable, fast and effective detection method. The method is expected to overcome the defects of the existing breast cancer diagnosis means and construct a more perfect and more advanced novel detection method.

Claims (2)

1. An ultra-wideband microwave early breast cancer detection device based on a support vector machine is characterized by comprising a feature extraction module, a data set construction module, a feature selection module and a support vector machine learning classification module, wherein the feature extraction module is used for acquiring an original signal from a breast to be detected and acquiring corresponding specific features; constructing an original characteristic data set through a data set construction module; optimizing the original characteristic data set through a characteristic selection module; finally, a support vector machine module is used for classifying the optimized data set to form an effective detection model, and a feature extraction module acquires signal features by combining a modal decomposition method with statistical characteristics; the characteristic selection module reduces the data dimension by projecting the characteristic data set to a subspace capable of keeping the maximum variance of the signal set by adopting a principal component analysis method and removes redundant characteristics; the detection method is characterized in that a Gaussian SVM method is used in a learning classification module of the support vector machine, echo signals obtained by detecting breasts according to an antenna array are detected by the following steps:
(1) carrying out normalization processing on echo signals of each antenna by using a minimum and maximum normalization algorithm;
(2) decomposing and analyzing a total M groups of echo signals acquired from the echo signals detected from a plurality of breasts by using a feature extraction module consisting of an integrated empirical mode decomposition, effective feature selection and statistical characteristic extraction method, and extracting all specific features of tumor signals and tumor-free signals;
(3) all the specific characteristics form an original characteristic data set through a data set construction module, the characteristics of the tumor echo signals are set as a positive sample, and a data label is set as 1; the characteristics of the tumor-free echo signals are set as negative samples, and the data label is set as-1;
(4) normalizing the data in the original characteristic data set again by using a minimum and maximum normalization algorithm to eliminate dimensional influences of different characteristics;
(5) in the feature selection module, a principal component analysis method is used for feature selection, M-dimensional data is changed to N-dimensional data through a corresponding algorithm, M is larger than N, and finally a simplified optimal data set with redundant variables removed is obtained;
(6) and (3) building and training a detection model on the optimized data set by using a support vector machine learning classification module based on a Gaussian kernel function.
2. The detection apparatus according to claim 1, wherein the feature extraction step of the feature extraction module comprises:
1) each 1 group of echo signals are decomposed into a plurality of groups of signal components through integrated empirical mode decomposition, and the component containing the most sufficient original signal information is judged and reserved through a correlation coefficient at the second stage;
2) taking 0.03 as a threshold value selected by the effective component, judging the IMF component with the correlation coefficient larger than the threshold value as the effective signal component, and discarding the rest signal components;
3) in the statistical characteristic extraction, the characteristic extraction based on 6 statistical characteristics of mean, variance, standard deviation, maximum value, energy entropy and information entropy is carried out on effective signal components, and specific characteristics are obtained from each 1 group of single-channel echo signals.
CN202010641919.4A 2020-07-06 2020-07-06 Ultra-wideband microwave early breast cancer detection device based on support vector machine Active CN111904417B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010641919.4A CN111904417B (en) 2020-07-06 2020-07-06 Ultra-wideband microwave early breast cancer detection device based on support vector machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010641919.4A CN111904417B (en) 2020-07-06 2020-07-06 Ultra-wideband microwave early breast cancer detection device based on support vector machine

Publications (2)

Publication Number Publication Date
CN111904417A CN111904417A (en) 2020-11-10
CN111904417B true CN111904417B (en) 2021-12-03

Family

ID=73227439

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010641919.4A Active CN111904417B (en) 2020-07-06 2020-07-06 Ultra-wideband microwave early breast cancer detection device based on support vector machine

Country Status (1)

Country Link
CN (1) CN111904417B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015164807A1 (en) * 2014-04-25 2015-10-29 Texas State University Detection of brain injury and subject state with eye movement biometrics
CN106023188A (en) * 2016-05-17 2016-10-12 天津大学 Breast tumor feature selection method based on Relief algorithm
CN106344006A (en) * 2016-11-03 2017-01-25 太原理工大学 J wave detection method based on pole symmetrical mode decomposition and support vector machine
CN107767946A (en) * 2017-09-26 2018-03-06 浙江工业大学 Breast cancer diagnosis system based on PCA and PSO KELM models
CN110500217A (en) * 2019-07-23 2019-11-26 南京航空航天大学 Based on can measured data feature common rail for diesel engine system oil pump fault detection method
CN110742606A (en) * 2019-10-12 2020-02-04 东华大学 Ultra-wideband microwave breast tumor size prediction system based on long-time and short-time memory

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7899625B2 (en) * 2006-07-27 2011-03-01 International Business Machines Corporation Method and system for robust classification strategy for cancer detection from mass spectrometry data

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015164807A1 (en) * 2014-04-25 2015-10-29 Texas State University Detection of brain injury and subject state with eye movement biometrics
CN106023188A (en) * 2016-05-17 2016-10-12 天津大学 Breast tumor feature selection method based on Relief algorithm
CN106344006A (en) * 2016-11-03 2017-01-25 太原理工大学 J wave detection method based on pole symmetrical mode decomposition and support vector machine
CN107767946A (en) * 2017-09-26 2018-03-06 浙江工业大学 Breast cancer diagnosis system based on PCA and PSO KELM models
CN110500217A (en) * 2019-07-23 2019-11-26 南京航空航天大学 Based on can measured data feature common rail for diesel engine system oil pump fault detection method
CN110742606A (en) * 2019-10-12 2020-02-04 东华大学 Ultra-wideband microwave breast tumor size prediction system based on long-time and short-time memory

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Analysis of tuberculosis disease through Raman spectroscopy and machine learning;Khan Saranjam等;《PHOTODIAGNOSIS AND PHOTODYNAMIC THERAPY》;20181022;第24卷;全文 *
基于EST和SVM的乳腺癌识别新方法;顾成扬等;《计算机工程与应用》;20110311(第08期);全文 *
基于PCA和SVM的管道腐蚀超声内检测;唐东林等;《传感技术学报》;20180715(第07期);全文 *

Also Published As

Publication number Publication date
CN111904417A (en) 2020-11-10

Similar Documents

Publication Publication Date Title
US10192099B2 (en) Systems and methods for automated screening and prognosis of cancer from whole-slide biopsy images
Milosevic et al. Thermography based breast cancer detection using texture features and minimum variance quantization
Chauhan et al. An efficient data mining classification approach for detecting lung cancer disease
US8139831B2 (en) System and method for unsupervised detection and gleason grading of prostate cancer whole mounts using NIR fluorscence
US20130208967A1 (en) Probability Density Function Estimation
US20160253466A1 (en) Systems and methods for quantitative analysis of histopathology images using multiclassifier ensemble schemes
Alqudah Ovarian cancer classification using serum proteomic profiling and wavelet features a comparison of machine learning and features selection algorithms
Goswami et al. Skin disease classification from image-a survey
Gordon et al. Image segmentation of uterine cervix images for indexing in PACS
WO2018157381A1 (en) Method and apparatus for intelligently classifying pathological slice image
Thilagaraj et al. Classification of breast cancer images by implementing improved dcnn with artificial fish school model
Djunaidi et al. Gray level co-occurrence matrix feature extraction and histogram in breast cancer classification with ultrasonographic imagery
Bhende et al. Attention layer-based multidimensional feature extraction for diagnosis of lung cancer
Asodekar et al. Brain tumor classification using shape analysis of MRI images
Wei et al. Multi-feature fusion for ultrasound breast image classification of benign and malignant
CN111904417B (en) Ultra-wideband microwave early breast cancer detection device based on support vector machine
Iftikhar et al. A colon cancer grade prediction model using texture and statistical features, smote and mrmr
Drzewiecki et al. Applicability of multifractal features as global characteristics of WorldView-2 panchromatic satellite images
Rozendo et al. Classification of non-Hodgkin lymphomas based on sample entropy signatures
Nyemeesha A systematic study and approach on detection of classification of skin cancer using back propagated artificial neural networks
Luo et al. Diagnosis of breast cancer tumor based on pca and fuzzy support vector machine classifier
Luo et al. Diagnosis of breast cancer tumor based on manifold learning and support vector machine
CN108960313B (en) Ultrasonic breast mass grading detection method based on Shearlet feature and hierarchical binary tree SVM classifier
Htay et al. Analysis on Results Comparison of Feature Extraction Methods for Breast Cancer Classification
Babu et al. HGMMEM Algorithm with Optimized DCNN Classification for Breast Cancer Image Screening and Detection

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
CP02 Change in the address of a patent holder

Address after: 300452 Binhai Industrial Research Institute Campus of Tianjin University, No. 48 Jialingjiang Road, Binhai New Area, Tianjin

Patentee after: Tianjin University

Address before: 300072 Tianjin City, Nankai District Wei Jin Road No. 92

Patentee before: Tianjin University

CP02 Change in the address of a patent holder