CN106845546A - A kind of breast X-ray image feature selection method based on BFBA and ELM - Google Patents

A kind of breast X-ray image feature selection method based on BFBA and ELM Download PDF

Info

Publication number
CN106845546A
CN106845546A CN201710048258.2A CN201710048258A CN106845546A CN 106845546 A CN106845546 A CN 106845546A CN 201710048258 A CN201710048258 A CN 201710048258A CN 106845546 A CN106845546 A CN 106845546A
Authority
CN
China
Prior art keywords
bat
represent
speed
new
stagnant
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
Application number
CN201710048258.2A
Other languages
Chinese (zh)
Other versions
CN106845546B (en
Inventor
韩晓红
相洁
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.)
Taiyuan University of Technology
Original Assignee
Taiyuan University of Technology
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 Taiyuan University of Technology filed Critical Taiyuan University of Technology
Priority to CN201710048258.2A priority Critical patent/CN106845546B/en
Publication of CN106845546A publication Critical patent/CN106845546A/en
Application granted granted Critical
Publication of CN106845546B publication Critical patent/CN106845546B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • 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
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Medical Informatics (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Image Analysis (AREA)

Abstract

A kind of breast X-ray image feature selection method based on BFBA and ELM of the present invention, it is related to technical field of image processing, technical problem to be solved is to avoid " index exploding " problem that dynamic programming runs into, and analytic method or bat algorithm are easily trapped into the problem of locally optimal solution；The technical scheme for using for：Comprise the following steps：The first step, collects data set MIAS used；Second step, sets BFBA parameters；3rd step, bat initialization of population；4th step, according to each bat corresponding character subset of position encoded generation；5th step, the search pulse frequency, speed and the position that update each bat；Six, the seven steps, generation uniform random number rand；8th step, the fitness value to all bats are ranked up, and find out current optimal solution and optimal value；9th step, judge whether optimal solution there occurs change；Tenth step, judge stagnant_count whether be equal to stagnant_max；11st step, repetition the 4th step to the tenth step；12nd step, output global optimum and optimal solution.

Description

A kind of breast X-ray image feature selection method based on BFBA and ELM
Technical field
The present invention relates to technical field of image processing, more particularly, to a kind of bat algorithm based on the response of birds colony The breast of (Bird Flock Bat Algorithm, BFBA) and extreme learning machine (Extreme Learning Machine, ELM) Gland radioscopic image feature selection approach.
Background technology
Mammary gland disease is one of common disease of women, while the multiple and harmfulness of breast cancer drastically influence women Health even life, therefore, the early diagnosis direct relation of mammary gland disease the health of human body of women.Particularly with mammary gland Cancer, present people can't completely determine its disease hair mechanism.The methods for clinical diagnosis of current breast cancer mainly includes touch examining Disconnected, histodiagnosis, cytodiagnosis and imaging diagnosis.It is convenience that iconography is diagnosed with it, scientific and of a relatively high Operability be widely adopted.Breast X-ray camera work is most common early diagnosing mammary cancer technology, for this side Method from breast X-ray image, it is necessary to analyze the disease condition of mammary gland.With the fast development of computer technology, mammary gland X is penetrated The analysis of line image also achieves the conversion that computer-assisted analysis is analyzed from Traditional Man, and this conversion can make breast cancer Diagnosis more rapidly, it is more accurate.
It is characterized in the key for determining similitude and classification.Due to recognizing breast cancer with computer, therefore from breast X-ray image On be extracted substantial amounts of primitive character to improve discrimination.But from extracting method, many features be not it is independent, i.e., this A little features have redundancy, and the speed and accuracy of Effect Mode identification, it is therefore desirable to select primitive character, are lost Those are equivocal, be difficult to differentiate or the strong feature of correlation.Feature selecting is substantially a combinatorial optimization problem.Conventional is excellent Change algorithm, such as analytic method can only obtain local optimum rather than globally optimal solution, and require that object function is continuous and can be micro-；Enumerate Although method overcomes these shortcomings, but computational efficiency is too low.Even if very famous dynamic programming, can also run into " index exploding " Problem, it for medium-scale and appropriate complexity problem, it is also usually helpless.Based on genetic algorithm, particle cluster algorithm Breast X-ray image feature selection method, obtain good classification results.Bat algorithm (Bat Algorithm, BA) is A kind of emerging heuristic Swarm Intelligence Algorithm, is compared to particle cluster algorithm and genetic algorithm, and BA can realize dynamic control Mutual transfer process between Local Search and global search, with the potential for playing a greater role.However, in BA search procedures In, there is the phenomenon of Premature Convergence, due to the rapid reduction of BA diversity of individuals, in order to obtain local optimum is more accurate Estimate, it is necessary to more invalid iteration.In this case, well balanced between being extremely difficult to explore and exploit, this makes Obtain BA and be easily stuck in local optimum.
The content of the invention
The present invention overcomes the shortcomings of that prior art is present, and technical problem to be solved is a kind of breast X-ray figure of offer As feature selection approach, it is to avoid " index exploding " problem that dynamic programming runs into, and analytic method or bat algorithm it is easy It is absorbed in the problem of locally optimal solution.
In order to solve the above-mentioned technical problem, the technical solution adopted by the present invention is：A kind of mammary gland X based on BFBA and ELM Ray image feature selection approach, comprises the following steps：
The first step, collects data set MIAS, i.e. the Mammographic Image Analysis Society used, Breast X-ray characteristics of image is extracted, and data set is divided into training set and test set, training set is used to train extreme learning machine ELM, i.e. Extreme Learning Machine, to design ELM graders, test set is used to check the effective of ELM graders Property；
The method for extracting the use of breast X-ray characteristics of image is gray level co-occurrence matrixes, extracts four kinds of statistical parameters：Angle two Rank square, entropy, the moment of inertia, coefficient correlation, the direction of gray level co-occurrence matrixes take 0 °, 45 °, 90 °, 135 ° of four directions；Calculate first Gray level co-occurrence matrixes on four direction, take pel spacing from being 1, secondly calculate four statistics ginsengs by each gray level co-occurrence matrixes Number；Each image is divided into four pieces, above-mentioned 16 features are extracted to every piece of subgraph as raw sample data, 64 are obtained altogether Statistical nature；Sample database obtained by feature extraction is divided into 10 parts, and selection wherein 90% is trained, and remaining 10% is tested；
Second step, sets BFBA parameters；
Initial parameter includes bat group size 20≤N≤100, each bat individual dimension D=64, pulse volume A =0.5, pulsation rate R=0.5, search pulse frequency range [Qmin,Qmax], wherein Qmin=0, Qmax=2, maximum iteration Iter_max=1000, stagnates counter stagnant_count=0, and maximum stagnates number of times stagnant_max=4；
3rd step, bat initialization of population, bat position vector is made up of 64 statistical natures that the first step is given, and uses Binary coded label feature group merges random initializtion bat body position and speed, forms sample set；
The position encoded use binary system of bat, primitive character has 64, i.e., individual length L=64, each individual Gene pairs answers the feature of corresponding order, i.e., when some gene in individuality is " 1 ", represent the corresponding characteristic item quilt of the gene From；Conversely, when being " 0 ", representing that this feature Xiang Wei is selected.Use X={ x1, x2, x3……xi……xNRepresent bat group The position of body, uses V={ v1, v2, v3……vi……vNRepresent speed, wherein xiRepresent i-th position of bat individuality, vi Represent i-th speed of bat individuality, xiAnd viIt is the row vector of 64 dimensions, a total N number of bat is individual, 20≤N≤100.
The formula that genic value is randomly generated is：
In formula, rand () is [0,1] interval independent identically distributed random number；
4th step, according to each bat corresponding character subset of position encoded generation, use this feature subset to generate training Collection and test set, training set are used to design ELM graders, and test set is used for testing classification device, corresponding according to test result calculations The fitness value fit of bati；
The calculating process of fitness value is as follows：
Wherein,
In formula, ωAThe accurate weights of formula presentation class, ωFRepresent feature selecting number weights, fjRepresent the characteristic value of gene：0 Or 1, acciPresentation class accuracy rate, cc represents correct classification number, and uc represents incorrect classification number；
5th step, the search pulse frequency, speed and the position that update each bat；
Search pulse frequency, speed and position are updated by below equation：
Qi=Qmin+(Qmax-Qmin)×β (11)
vi(t)=vi(t-1)+(xi(t)-xbest)×Qi (12)
xi(t)=xi(t-1)+vi(t) (13)
In formula：β belongs to [0,1], is equally distributed random number；QiIt is the search pulse frequency of bat i, QiBelong to [Qmin, Qmax]；vi(t)、vi(t-1) speed of the bat i at t the and t-1 moment is represented respectively；xi(t)、xi(t-1) represent that bat i exists respectively The position at t and t-1 moment；xbestRepresent the optimal solution of current all bats；
6th step, generation uniform random number rand, if rand>R, R are the pulsation rate described in second step, then right Current optimal solution carries out random perturbation, produces a new solution：
xnew=xbest+ 0.001 × randn (1,64)； (14)
Wherein, xnewRepresent the new new explanation for producing, xbestRepresent optimal solution at that time.
7th step, generation uniform random number rand, if rand meets rand<A and fit (xi(t))<fit(xi(t- 1)), A is the pulse volume described in second step, then receive the new explanation of step 6 generation：
xi(t)=xnew (15)
Wherein, fit (xi(t)) represent and seek individual xiFitness value, xnewRepresent the new explanation that the 6th step is produced；
If rand does not meet above-mentioned condition, this step is skipped, into next step；
8th step, the fitness value to all bats are ranked up, and find out current optimal solution and optimal value；
9th step, judge whether optimal solution there occurs change, if do not changed,
Stagnant_count=stagnant_count+1, otherwise, stagnant_count=0
Wherein stagnant_count is the stagnation counter described in second step；
Tenth step, judge whether stagnant_count is equal to stagnant_max, if equal, all of bat hair Raw colony's response；
If i.e. by the iteration of certain number of times, optimal overall situation adaptive value no longer changes, then birds colony occurs and rings The repositioning answered, speed and the position of each bat are updated using formula (16) and (17) again, in formula (16), It is batNew position after being repositioned by colony's response, new position is by calculating its seven nearest objects Position mean is obtained,Nearest bat beWith the minimum Eustachian distance of other bats, randxIt is interval [- 1,1] A random real number, in formula (17),It is new speed of the bat after speed is adjusted,It is the original speed of bat, New speed is obtained by calculating the speed average of its seven nearest bats, randvIt is that [0,1] interval one is random Real number, NiComprisingSeven index values of neighbouring bat；
11st step, repetition the 4th step to the tenth step are until meeting the optimal solution conditions of setting or reaching greatest iteration time Number；
12nd step, output global optimum and optimal solution.
In the first step, data set MIAS is the standard data set for studying breast X-ray image, image is 1024 × 1024 gray-scale map.
The present invention has the advantages that compared with prior art.
The inventive method realizes the optimum choice to breast X-ray image high dimensional feature, effectively reduces intrinsic dimensionality, carries The degree of accuracy of high-class identification, and the requirement of extreme learning machine grader is adapted to, further improve classification performance.The method is special Levy Selection effect good, efficiency high can effectively improve the nicety of grading of breast X-ray image.
BFBA introduces a new diversity exploration mechanism and explores ability, the machine to strengthen its diversity in the present invention System comes from colony's respondent behavior of birds.By colony's response mechanism, BFBA can explore the wider array of search space of scope, and because This is avoided limiting into local sub-optimum solution.ELM is calculated by a step and can be solved the output weights for separating out learning network, with god Compared with SVMs through network, extreme learning machine drastically increases the generalization ability and pace of learning of network, the present invention By the use of ELM as the importance of classifier evaluation feature.
Brief description of the drawings
The present invention will be further described in detail below in conjunction with the accompanying drawings.
Fig. 1 is certain individual gene code in the present invention.
Fig. 2 is flow chart of the invention.
Specific embodiment
As shown in Figure 1 and Figure 2, a kind of breast X-ray image feature selection method based on BFBA and ELM of the present invention, specifically Step is as follows：The first step, collects data set MIAS, i.e. the Mammographic Image Analysis Society used, Breast X-ray characteristics of image is extracted, and data set is divided into training set and test set, training set is used to train extreme learning machine ELM, i.e. Extreme Learning Machine, to design ELM graders, test set is used to check the effective of ELM graders Property；
The method for extracting the use of breast X-ray characteristics of image is gray level co-occurrence matrixes, extracts four kinds of statistical parameters：Angle two Rank square, entropy, the moment of inertia, coefficient correlation, the direction of gray level co-occurrence matrixes take 0 °, 45 °, 90 °, 135 ° of four directions；Calculate first Gray level co-occurrence matrixes on four direction, take pel spacing from being 1, secondly calculate four statistics ginsengs by each gray level co-occurrence matrixes Number；Each image is divided into four pieces, above-mentioned 16 features are extracted to every piece of subgraph as raw sample data, 64 are obtained altogether Statistical nature；Sample database obtained by feature extraction is divided into 10 parts, and selection wherein 90% is trained, and remaining 10% is tested；
Angular second moment f1Represent, entropy f2Expression, the moment of inertia f3Expression, coefficient correlation f4Represent：
In formula, μ1, μ2,WithIt is respectively defined as：
Feature database is randomly divided into 10 parts, and selection wherein 90% is trained, and remaining 10% is tested；
Second step, sets BFBA parameters；
Initial parameter includes bat group size 20≤N≤100, each bat individual dimension D=64, pulse volume A =0.5, pulsation rate R=0.5, search pulse frequency range [Qmin,Qmax], wherein Qmin=0, Qmax=2, maximum iteration Iter_max=1000, stagnates counter stagnant_count=0, and maximum stagnates number of times stagnant_max=4；
3rd step, bat initialization of population, bat position vector is made up of 64 statistical natures that the first step is given, and uses Binary coded label feature group merges random initializtion bat body position and speed, forms sample set；
The position encoded use binary system of bat, primitive character has 64, i.e., individual length L=64, each individual Gene pairs answers the feature of corresponding order, i.e., when some gene in individuality is " 1 ", represent the corresponding characteristic item quilt of the gene From；Conversely, when being " 0 ", representing that this feature Xiang Wei is selected.Use X={ x1, x2, x3……xi……xNRepresent bat group The position of body, uses V={ v1, v2, v3……vi……vNRepresent speed, wherein xiRepresent i-th position of bat individuality, vi Represent i-th speed of bat individuality, xiAnd viIt is the row vector of 64 dimensions, a total N number of bat is individual, 20≤N≤100.
The formula that genic value is randomly generated is：
In formula, rand () is [0,1] interval independent identically distributed random number；
4th step, according to each bat corresponding character subset of position encoded generation, use this feature subset to generate training Collection and test set, training set are used to design ELM graders, and test set is used for testing classification device, corresponding according to test result calculations The fitness value fit of bati；
The calculating process of fitness value is as follows：
Wherein,
In formula, ωAThe accurate weights of formula presentation class, ωFRepresent feature selecting number weights, fjRepresent the characteristic value of gene：0 Or 1, acciPresentation class accuracy rate, cc represents correct classification number, and uc represents incorrect classification number；
5th step, the search pulse frequency, speed and the position that update each bat；
Search pulse frequency, speed and position are updated by below equation：
Qi=Qmin+(Qmax-Qmin)×β (11)
vi(t)=vi(t-1)+(xi(t)-xbest)×Qi (12)
xi(t)=xi(t-1)+vi(t) (13)
In formula：β belongs to [0,1], is equally distributed random number；QiIt is the search pulse frequency of bat i, QiBelong to [Qmin, Qmax]；vi(t)、vi(t-1) speed of the bat i at t the and t-1 moment is represented respectively；xi(t)、xi(t-1) represent that bat i exists respectively The position at t and t-1 moment；xbestRepresent the optimal solution of current all bats；
6th step, generation uniform random number rand, if rand>R, R are the pulsation rate described in second step, then right Current optimal solution carries out random perturbation, produces a new solution：
xnew=xbest+ 0.001 × randn (1,64)； (14)
Wherein, xnewRepresent the new new explanation for producing, xbestRepresent optimal solution at that time.
7th step, generation uniform random number rand, if rand meets rand<A and fit (xi(t))<fit(xi(t- 1)), A is the pulse volume described in second step, then receive the new explanation of step 6 generation：
xi(t)=xnew (15)
Wherein, fit (xi(t)) represent and seek individual xiFitness value, xnewRepresent the new explanation that the 6th step is produced；
If rand does not meet above-mentioned condition, this step is skipped, into next step；
8th step, the fitness value to all bats are ranked up, and find out current optimal solution and optimal value；
9th step, judge whether optimal solution there occurs change, if do not changed,
Stagnant_count=stagnant_count+1, otherwise, stagnant_count=0
Wherein stagnant_count is the stagnation counter described in second step；
Tenth step, judge whether stagnant_count is equal to stagnant_max, if equal, all of bat hair Raw colony's response；
If i.e. by the iteration of certain number of times, optimal overall situation adaptive value no longer changes, then birds colony occurs and rings The repositioning answered, speed and the position of each bat are updated using formula (16) and (17) again, in formula (16), It is batNew position after being repositioned by colony's response, new position is by calculating its seven nearest objects Position mean is obtained,Nearest bat beWith the minimum Eustachian distance of other bats, randxIt is interval [- 1,1] A random real number, in formula (17),It is new speed of the bat after speed is adjusted,It is the original speed of bat, New speed is obtained by calculating the speed average of its seven nearest bats, randvIt is that [0,1] interval one is random Real number, NiComprisingSeven index values of neighbouring bat；
11st step, repetition the 4th step to the tenth step are until meeting the optimal solution conditions of setting or reaching greatest iteration time Number；
12nd step, output global optimum and optimal solution.
In the first step, data set MIAS is the standard data set for studying breast X-ray image, image is 1024 × 1024 gray-scale map.
The present invention can be preferably analyzed breast X-ray image, can more accurately analyze the ill feelings of mammary gland Condition, helps doctor more accurately and rapidly to diagnose and formulate therapeutic scheme.

Claims (2)

1. a kind of breast X-ray image feature selection method based on BFBA and ELM, it is characterised in that comprise the following steps：
The first step, collects data set MIAS, i.e. the Mammographic Image Analysis Society used, extracts Breast X-ray characteristics of image, and data set is divided into training set and test set, training set is used to train extreme learning machine ELM, i.e., Extreme Learning Machine, to design ELM graders, test set is used to check the validity of ELM graders；
The method for extracting the use of breast X-ray characteristics of image is gray level co-occurrence matrixes, extracts four kinds of statistical parameters：Angle second order Square, entropy, the moment of inertia, coefficient correlation, the direction of gray level co-occurrence matrixes take 0 °, 45 °, 90 °, 135 ° of four directions；Four are calculated first Gray level co-occurrence matrixes on individual direction, take pel spacing from being 1, secondly calculate four statistics ginsengs by each gray level co-occurrence matrixes Number；Each image is divided into four pieces, above-mentioned 16 features are extracted to every piece of subgraph as raw sample data, 64 are obtained altogether Statistical nature；Sample database obtained by feature extraction is divided into 10 parts, and selection wherein 90% is trained, and remaining 10% is tested；
Second step, sets BFBA parameters；
Initial parameter includes bat group size 20≤N≤100, each bat individual dimension D=64, pulse volume A= 0.5, pulsation rate R=0.5, search pulse frequency range [Qmin,Qmax], wherein Qmin=0, Qmax=2, maximum iteration Iter_max=1000, stagnates counter stagnant_count=0, and maximum stagnates number of times stagnant_max=4；
3rd step, bat initialization of population, bat position vector is made up of 64 statistical natures that the first step is given, and enters using two Coded markings feature group processed merges random initializtion bat body position and speed, forms sample set；
The position encoded use binary system of bat, primitive character has 64, i.e., individual length L=64, each individual gene The feature of the corresponding order of correspondence, i.e., when some gene in individuality is " 1 ", represent that the corresponding characteristic item of the gene is chosen With；Conversely, when being " 0 ", representing that this feature Xiang Wei is selected.Use X={ x1, x2, x3……xi……xNRepresent bat colony Position, use V={ v1, v2, v3……vi……vNRepresent speed, wherein xiRepresent i-th position of bat individuality, viTable Show i-th speed of bat individuality, xiAnd viIt is the row vector of 64 dimensions, a total N number of bat is individual, 20≤N≤100.
The formula that genic value is randomly generated is：
x i = 0 r a n d ( ) < 0.5 1 r a n d ( ) &GreaterEqual; 0.5 - - - ( 9 )
In formula, rand () is [0,1] interval independent identically distributed random number；
4th step, according to each bat corresponding character subset of position encoded generation, using this feature subset generate training set and Test set, training set is used to design ELM graders, and test set is used for testing classification device, according to the corresponding bat of test result calculations Fitness value fiti；
The calculating process of fitness value is as follows：
fit i = acc i &times; &omega; A + &lsqb; 1 - &Sigma; j = 1 L f j L &rsqb; &times; &omega; F - - - ( 10 )
Wherein,
In formula, ωAThe accurate weights of formula presentation class, ωFRepresent feature selecting number weights, fjRepresent the characteristic value of gene：0 or 1, acciPresentation class accuracy rate, cc represents correct classification number, and uc represents incorrect classification number；
5th step, the search pulse frequency, speed and the position that update each bat；
Search pulse frequency, speed and position are updated by below equation：
Qi=Qmin+(Qmax-Qmin)×β (11)
vi(t)=vi(t-1)+(xi(t)-xbest)×Qi (12)
xi(t)=xi(t-1)+vi(t) (13)
In formula：β belongs to [0,1], is equally distributed random number；QiIt is the search pulse frequency of bat i, QiBelong to [Qmin, Qmax]；vi(t)、vi(t-1) speed of the bat i at t the and t-1 moment is represented respectively；xi(t)、xi(t-1) represent that bat i exists respectively The position at t and t-1 moment；xbestRepresent the optimal solution of current all bats；
6th step, generation uniform random number rand, if rand>R, R are the pulsation rate described in second step, then to current Optimal solution carries out random perturbation, produces a new solution：
xnew=xbest+ 0.001 × randn (1,64)； (14)
Wherein, xnewRepresent the new new explanation for producing, xbestRepresent optimal solution at that time.
7th step, generation uniform random number rand, if rand meets rand<A and fit (xi(t))<fit(xi(t-1)), A is the pulse volume described in second step, then receive the new explanation of step 6 generation：
xi(t)=xnew (15)
Wherein, fit (xi(t)) represent and seek individual xiFitness value, xnewRepresent the new explanation that the 6th step is produced；
If rand does not meet above-mentioned condition, this step is skipped, into next step；
8th step, the fitness value to all bats are ranked up, and find out current optimal solution and optimal value；
9th step, judge whether optimal solution there occurs change, if do not changed,
Stagnant_count=stagnant_count+1, otherwise, stagnant_count=0
Wherein stagnant_count is the stagnation counter described in second step；
Tenth step, judge whether stagnant_count is equal to stagnant_max, if equal, all of bat occurs group Body is responded；
If i.e. by the iteration of certain number of times, optimal overall situation adaptive value no longer changes, then the response of birds colony occurs Reposition, update speed and the position of each bat again using formula (16) and (17), in formula (16),It is bat BatNew position after being repositioned by colony's response, new position is by calculating its seven positions of nearest object Average value is obtained,Nearest bat beWith the minimum Eustachian distance of other bats, randxIt is the one of [- 1,1] interval Individual random real number, in formula (17),It is new speed of the bat after speed is adjusted,It is the original speed of bat, new speed Spend the speed average by calculating its seven nearest bats to obtain, randvIt is interval random reality of [0,1] Number, NiComprisingSeven index values of neighbouring bat；
11st step, repetition the 4th step to the tenth step are until meeting the optimal solution conditions of setting or reaching maximum iteration；
12nd step, output global optimum and optimal solution.
2. a kind of breast X-ray image feature selection method based on BFBA and ELM according to claim 1, its feature It is：In the first step, data set MIAS is the standard data set for studying breast X-ray image, and image is 1024 × 1024 Gray-scale map.
CN201710048258.2A 2017-01-20 2017-01-20 BFBA and ELM-based mammary X-ray image feature selection method Expired - Fee Related CN106845546B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710048258.2A CN106845546B (en) 2017-01-20 2017-01-20 BFBA and ELM-based mammary X-ray image feature selection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710048258.2A CN106845546B (en) 2017-01-20 2017-01-20 BFBA and ELM-based mammary X-ray image feature selection method

Publications (2)

Publication Number Publication Date
CN106845546A true CN106845546A (en) 2017-06-13
CN106845546B CN106845546B (en) 2020-09-04

Family

ID=59119468

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710048258.2A Expired - Fee Related CN106845546B (en) 2017-01-20 2017-01-20 BFBA and ELM-based mammary X-ray image feature selection method

Country Status (1)

Country Link
CN (1) CN106845546B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112107332A (en) * 2020-10-10 2020-12-22 高慧强 Method, equipment and system for processing medical ultrasonic image

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103593674A (en) * 2013-11-19 2014-02-19 太原理工大学 Cervical lymph node ultrasonoscopy feature selection method
US20160292509A1 (en) * 2010-08-26 2016-10-06 Blast Motion Inc. Sensor and media event detection and tagging system
CN106203614A (en) * 2016-07-22 2016-12-07 吉林大学 KP model densities Function identification method based on self adaptation Vespertilio searching algorithm

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160292509A1 (en) * 2010-08-26 2016-10-06 Blast Motion Inc. Sensor and media event detection and tagging system
CN103593674A (en) * 2013-11-19 2014-02-19 太原理工大学 Cervical lymph node ultrasonoscopy feature selection method
CN106203614A (en) * 2016-07-22 2016-12-07 吉林大学 KP model densities Function identification method based on self adaptation Vespertilio searching algorithm

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
DORESWAMY ET AL.: "BAT-ELM: A Bio Inspired Model for Prediction of Breast Cancer Data", 《2015 INTERNATIONAL CONFERENCE ON APPLIED AND THEORETICAL COMPUTING AND COMMUNICATION TECHNOLOGY》 *
JIE XIANG ET AL.: "A novel hybrid system for feature selection based on an improved gravitational search algorithm and k-NN method", 《APPLIED SOFT COMPUTING》 *
冷令: "蝙蝠算法联合选择特征和分类器参数的入侵检测", 《计算机应用与软件》 *
李斌 等: "基于BGSA和ELM的垃圾邮件检测", 《山西科技》 *
胡楠 等: "基于鸟类群体响应的万有引力搜索算法及其在数据聚类中的应用", 《陕西科技》 *
陈海挺: "改进蝙蝠算法优化极限学习机的图像分类", 《激光杂志》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112107332A (en) * 2020-10-10 2020-12-22 高慧强 Method, equipment and system for processing medical ultrasonic image

Also Published As

Publication number Publication date
CN106845546B (en) 2020-09-04

Similar Documents

Publication Publication Date Title
Singh et al. Screening of COVID-19 suspected subjects using multi-crossover genetic algorithm based dense convolutional neural network
CN102395975B (en) Clinical Decision Support Systems and method
SR et al. Lung cancer detection using probabilistic neural network with modified crow-search algorithm
Khan et al. Beyond activity recognition: skill assessment from accelerometer data
CN106778042A (en) Cardio-cerebral vascular disease patient similarity analysis method and system
CN103593674B (en) A kind of cervical lymph node ultrasonoscopy feature selection method
CN112472048B (en) Method for realizing neural network for identifying pulse condition of cardiovascular disease patient
CN110335668A (en) Thyroid cancer cell pathological map auxiliary analysis method and system based on deep learning
Fan et al. Evolutionary neural architecture search for retinal vessel segmentation
CN110188654A (en) A kind of video behavior recognition methods not cutting network based on movement
CN116110597B (en) Digital twinning-based intelligent analysis method and device for patient disease categories
Deshpande et al. Improved Otsu and Kapur approach for white blood cells segmentation based on LebTLBO optimization for the detection of Leukemia.
CN110720935A (en) Multidimensional attention concentration capability evaluation method
Madhu et al. Imperative dynamic routing between capsules network for malaria classification
CN113397485A (en) Scoliosis screening method based on deep learning
CN113782184A (en) Cerebral apoplexy auxiliary evaluation system based on facial key point and feature pre-learning
CN113095382A (en) Interpretable tuberculosis classification network identification method based on CT image
Bandyopadhyay et al. Machine learning and deep learning integration for skin diseases prediction
CN115564756A (en) Medical image focus positioning display method and system
CN115471885A (en) Action unit correlation learning method and device, electronic device and storage medium
Naskinova Transfer learning with NASNet-Mobile for Pneumonia X-ray classification
Li et al. Explainable human‐in‐the‐loop healthcare image information quality assessment and selection
CN113052227A (en) Pulmonary tuberculosis identification method based on SE-ResNet
CN110378353A (en) A kind of tongue picture feature extracting method, system and computer readable storage medium
CN106845546A (en) A kind of breast X-ray image feature selection method based on BFBA and ELM

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
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20200904