CN116596836A - Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy - Google Patents

Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy Download PDF

Info

Publication number
CN116596836A
CN116596836A CN202310226858.9A CN202310226858A CN116596836A CN 116596836 A CN116596836 A CN 116596836A CN 202310226858 A CN202310226858 A CN 202310226858A CN 116596836 A CN116596836 A CN 116596836A
Authority
CN
China
Prior art keywords
pneumonia
sample
neighborhood
attribute
image data
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
Application number
CN202310226858.9A
Other languages
Chinese (zh)
Inventor
鞠恒荣
樊晓雪
丁卫平
黄嘉爽
曹金鑫
程纯
姜舒
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.)
Nantong University
Original Assignee
Nantong 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 Nantong University filed Critical Nantong University
Priority to CN202310226858.9A priority Critical patent/CN116596836A/en
Publication of CN116596836A publication Critical patent/CN116596836A/en
Pending legal-status Critical Current

Links

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
    • 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
    • G06V10/513—Sparse representations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30061—Lung
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Databases & Information Systems (AREA)
  • Public Health (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Data Mining & Analysis (AREA)
  • Pathology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

The invention provides a multi-view neighborhood evidence entropy-based pneumonia CT image attribute reduction method, belongs to the technical field of intelligent medical treatment, and solves the problem that pathological conditions cannot be detected in time due to excessive redundant pathological attributes in a pneumonia CT image. The technical proposal is as follows: the method comprises the following steps: s10, reading a pneumonia CT image data set; s20, representing the correlation between the CT image data samples of the pneumonia through two views of sparse representation and distance measurement, and describing a neighborhood range through two stages of fusion and deletion; s30, fusing the evidence theory and the neighborhood entropy to construct a neighborhood evidence entropy for describing the attribute importance of the CT image data of the pneumonia; s40, calculating an attribute reduction set in the CT image dataset of the pneumonia. The beneficial effects of the invention are as follows: the detection time of the CT image data of the pneumonia is reduced, the detection efficiency is improved, and a doctor is helped to effectively analyze the pathological change condition of the pneumonia.

Description

Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy
Technical Field
The invention relates to the technical field of intelligent medical treatment, in particular to a pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy.
Background
Pneumonia is caused by the invasion of pathogens into the lung parenchyma and overgrowth in the lung parenchyma beyond the host's defenses, resulting in the appearance of exudates in the alveolar space, which causes pneumonia. The occurrence and severity of pneumonia is primarily determined by a balance between pathogen factors and host factors. The common symptoms of the cough and the expectoration are cough, thick phlegm or blood phlegm, and dyspnea and the like are accompanied. The symptoms are more atypical, the disease progress is rapid, and missed diagnosis and misdiagnosis are easy to occur.
With the continuous development of computer technology, automatic detection of the CT image of pneumonia is realized by using artificial intelligence so as to ensure the correctness of the diagnosis result. At present, students at home and abroad also conduct intensive research on the detection, and a plurality of better detection models are provided. When facing a wide range of data information, computer technology can effectively process, delete redundant information and extract effective information, thereby continuously improving detection efficiency. However, the method proposed at present does not effectively fuse the data information, but only considers the relation existing between the data from a single angle, so how to effectively fuse the data information is one of the problems to be considered in the invention.
In addition, the effective method for judging the pathological condition of the pneumonia is to pass through a pathological attribute experiment of the pathogenesis of the pneumonia, however, in the experimental process, excessive pathological attribute data is needed, and a new method is needed to effectively reduce redundant attribute in the classification information of the CT image data of the pneumonia, reduce the detection time of the CT image data of the pneumonia, improve the detection efficiency and help doctors to effectively analyze the pathological condition of the pneumonia.
How to solve the technical problems is the subject of the present invention.
Disclosure of Invention
In order to solve the problems, the invention provides a method for reducing the attributes of the CT images of the pneumonia based on multi-view neighborhood evidence entropy, which is used for describing a neighborhood set of the CT images of the pneumonia from two views of distance measurement and sparse representation, introducing neighborhood entropy and evidence theory, constructing neighborhood evidence entropy, describing the importance of each attribute, removing redundant attributes, effectively improving the efficiency and the precision of the attribute reduction of the CT images of the pneumonia and having stronger application value for intelligent auxiliary diagnosis of the pneumonia.
In order to achieve the above purpose, the invention adopts the following technical scheme: a pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy comprises the following steps:
s10, reading a pneumonia CT image data set, and converting the data set into a four-tuple decision information system;
s20, representing the correlation between the CT image data samples of the pneumonia through two views of sparse representation and distance measurement, and describing a neighborhood range through two stages of fusion and deletion;
s30, fusing the evidence theory and the neighborhood entropy to construct a neighborhood evidence entropy for describing the attribute importance of the CT image data of the pneumonia;
s40, calculating an attribute reduction set in the CT image dataset of the pneumonia.
Further, the step S10 includes the steps of:
s11, reading a pneumonia CT image data set, determining an attribute set and decision classification of the pneumonia CT image data set, wherein the decision information system S=<U,C∪D,V,f>Wherein u= { x 1 ,…,x i ,…,x N Sample set of CT image data of pneumonia, N represents the number of CT images of pneumonia, x i Represents the ith sample, x N Representing the nth sample; c= { a 1 ,…,a i ,…,a n CT image of pneumoniaA non-empty finite set of data attributes, n representing the number of attributes in the CT image data of pneumonia, a i Represents the ith attribute, a n Represents an nth attribute; d= { D 1 ,…,d i ,…,d m The non-empty finite set of decision attributes of the CT image data of the pneumonia is represented, m represents the number of decision categories in the CT image data of the pneumonia, and d i Represents the i-th decision category, d m Representing an mth decision category; v= U-shaped gate a∈C∪D V a ，V a Is the possible case of the CT image data attribute a of pneumonia; u x (C U D) V is an information function which assigns an information value to each CT image, i.ex∈U,f(x,a)∈V a ；
S12, dividing the pneumonia CT image data set S into t data subsets according to the number of different information values of the decision attribute D in the data set, and meeting the requirements of wherein Si Represents the ith subset of data, S j Represents the j-th subset of data, i+.j, i=1, 2, …, t, j=1, 2, …, t.
Further, the step S20 includes the steps of:
s21, setting a pneumonia CT image data set as X, setting a reconstruction weight matrix of X as W, and calculating a sample X to be inspected through a sparse constraint function i And other samples { x } 1 ,x 2 ,…,x i-1 ,x i+1 ,…,x N Sparse correlation between samples x to be examined i The objective function of (2) is as follows:
wherein ,xi-1 Represents the i-1 th sample, x i+1 Representing the (i + 1) th sample,is the square of the Frobenius norm, F is a selection mode of norms, I.I.I.I 1 Is a 1-norm, T is a transposed transform of the matrix, η and μ are parameters having values between 0 and 1, R (W) =Tr (W T X T LXW) is a locally preserved projection regularization term, where L is a laplace matrix representing the relationship information between attributes, and Tr is the trace of the matrix, i.e., the sum of the diagonal elements of the matrix.
S22, according to a weight matrix W obtained by reconstructing the CT image data set of the pneumonia, sparse correlation among samples can be seen, if the sparse correlation is larger than 0, the two samples are considered to be in positive correlation, the two samples are put into a sparse neighborhood set, and the number of the samples in positive correlation is recorded as k. For the following Sample x i Sparse neighborhood set SN derived from sparse representation A (x i ) The definition is as follows:
SN A (x i )＝{x j |x j ∈U,W(x i ,x j )＞0}
wherein A represents a subset of attributes in a non-null finite set of data attributes C, x j Represents the jth sample, W (x i ,x j ) Represents x i and xj Sparse correlation weight values between.
S23, calculating CT image sample x of pneumonia i And other samples x j Between Euclidean distance dis (x) i ,x j ) It is defined as follows:
wherein f (x, a) t ) Representing sample x at attribute a of t t Is a value of (b).
S24, according to Euclidean distance and k value, adopting k neighbor thought, namely selecting the first k neighbors according to the distance, and solvingA set of distance neighbors. For the following Sample x i Distance neighborhood set KN from k neighbors A (x i ) The definition is as follows:
wherein ,representing distance x i N-th sample of>Represents the distance x i The distance of the nearest nth sample.
S25, fusing the sparse neighborhood set and the distance neighborhood set of the CT image data of the pneumonia to obtain a multi-view neighborhood set MN of the CT image data set of the pneumonia A (x i ) The method comprises the following steps:
MN A (x i )＝{x j |x j ∈SN A (x i )∪x j ∈KN A (x i )}
s26, performing first-order deletion operation on the fused neighborhood set, and calculating a sample x in the CT (computed tomography) image data set i Sample x in its neighborhood set j The average distance of (2) is denoted ave. Comparing dis (x) i ,x j ) And ave, if dis (x i ,x j )>ave, consider sample x j And sample x i The first order neighborhood relation is not satisfied, and the sample x is obtained j From x i Is deleted from the multi-view neighborhood set. For the followingSample x i Through the firstThe first-order neighborhood set obtained after one deletion isThe calculation formula is as follows:
and S27, performing second-order operation on a first-order neighborhood set of samples in the CT image data set of the pneumonia according to the midline theorem, and deleting samples which do not meet the second-order requirement. Determining sample x in a neighborhood set using a perpendicular bisector j Whether or not the sample x exists on the perpendicular bisector i Is provided. Count all samples in the neighborhood set if sample x j Sample x to be examined i If the value is not changed, otherwise the value is increased by 1. After counting, consider the sample with the largest value as the sample x to be examined i The two-order neighborhood relation is not satisfied, and the two-order neighborhood relation is regarded as a redundant sample. And deleting the redundant samples to obtain a second-order neighborhood set of the CT image data samples of the pneumonia. According to the theorem of the perpendicular bisectorSample x i The second order neighborhood set obtained after the second deletion is +.>The definition of puncturing is as follows:
wherein count represents the calculation of the sample x from the perpendicular bisector j Whether to be examined sample x i The values on the same side, n', represent the sample x to be examined i Is the number in the first order neighborhood set,representing the calculation of x from the perpendicular bisector i Samples in the first order neighborhood are not in sample x to be examined i Maximum on the same side.
Further, the step S30 includes the steps of:
s31, calculating a p-th sample x to be examined in a pneumonia CT image data set p And other sample x of q q Distance dis (x) p ,x q ) And sparse representation W (x p ,x q ) Respectively denoted as dis p,q and spap,q According to dis p,q and spap,q Calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)
S32, calculating other samples x in the CT image dataset of the pneumonia q For the sample x to be examined p Evidence information provided, Ω=u/d= { D 1 ,D 2 ,...,D j The decision class division set of the CT image dataset of the pneumonia is shown as the figure of j, D j For the j decision class division, if the CT image sample x of the pneumonia q Class label of l t And decision class division into D t Then (x) q ,D t ) Can be independently the sample x to be examined p Providing evidence support, including evidence informationThe definition is as follows:
wherein ,Dt For the t decision class division, w 1 and w2 Importance weight duty cycle representing distance measure and sparse representation, w 1 +w 2 ＝1. μ is a correlation parameter of the distance metric, η is a correlation parameter of the sparse representation, and μ > 0 and η > 0 are satisfied.
S33, calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)The middle class label is l t Is as followsSample set->For x p Evidence support of (a) can be expressed as +.>Namely:
s34, calculating CT image sample x of pneumonia p Evidence information under different category labelsAnd performs a merging operation on the certification information, which may be specifically expressed as:
wherein ,j is the number of decision divisions.
S35, calculating CT image sample x of pneumonia p Dividing decision classes Ω=u/d= { D 1 ,D 2 ,...,D j D in } t Trust function for classThe specific definition is as follows:
s36, calculating CT image sample x of pneumonia p Neighborhood evidence entropy H (d|a), x p Is set as the second order neighborhood of (2)[x p ] D Is x p Is defined as follows:
wherein, |U| is the number of domains.
Further, the step S40 includes the steps of:
s41, calculating each pathological attribute in the CT image dataset of the pneumoniaExternal importance measure sig with respect to reduced set red out (red, a, D) =h (d|red) -H (d|red u { a }) and selects the attribute a corresponding to the greatest attribute importance k ；
S42, judging sig out (red,a k D) if it is greater than 0, if sig is satisfied out (red,a k D) > 0, attribute a will be k Adding the candidate attribute subset C into the reduced set red, and deleting the candidate attribute subset C;
s43, calculating and comparing H (D|red) andh (D|C), if H (D|red) > H (D|C) and sig out (red,a k D) is less than or equal to 0, attribute a k Delete from reduced set red until
S44, for each attribute in the reduced set redRe-calculating the importance measure, sig, of the attribute in (red,a w ,D)＝H(D|red-{a w -H (d|red), deleting attribute a in turn w Comparison of sig in (red,a w D) if it is greater than 0, if sig in (red,a w D) is less than or equal to 0, attribute a w Deleted from the reduced set red. And (3) after sequentially circularly traversing all the attributes in red, outputting a reduced set red of the pneumonia CT image data set.
Compared with the prior art, the invention has the beneficial effects that:
(1) According to the pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy, the neighborhood of the pneumonia CT image is firstly described from two views of sparse representation and distance measurement, the neighborhood is described more accurately in a multi-view mode, wherein the k is the number of the neighborhood of certain pneumonia CT image data by using a self-adaptive k value, the definition parameters are not needed to be manually carried out, and errors are reduced to a certain extent.
(2) Introducing evidence theory into the neighborhood entropy, and putting the trust function into the neighborhood entropy as support information between samples to judge the uncertainty of the attribute. Finally, heuristic reduction Jian Suanfa is designed according to the neighborhood evidence entropy, and the reduced attributes are selected. The method improves the efficiency and the precision of the attribute reduction of the CT image data of the pneumonia and improves the application value of intelligent auxiliary diagnosis of the pneumonia.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Fig. 1 is a general flowchart of a method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy according to the present invention.
Fig. 2 is a diagram of an overall data processing framework of the method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy.
Fig. 3 is a flow chart of attribute reduction of the method for reducing the attributes of the CT image of the pneumonia based on multi-view neighborhood evidence entropy.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. Of course, the specific embodiments described herein are for purposes of illustration only and are not intended to limit the invention.
Example 1
Referring to fig. 1 to 3, the present embodiment provides a method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy, which includes the following steps:
s10, reading a pneumonia CT image data set, and converting the data set into a four-tuple decision information system;
s20, representing the correlation between the CT image data samples of the pneumonia through two views of sparse representation and distance measurement, and describing a neighborhood range through two stages of fusion and deletion;
s30, fusing the evidence theory and the neighborhood entropy, constructing a neighborhood evidence entropy, and describing attribute importance of the CT image data of the pneumonia;
s40, calculating an attribute reduction set in the CT image data set of the pneumonia.
The step S10 includes the steps of:
s11, reading a pneumonia CT image data set, determining an attribute set and decision classification of the pneumonia CT image data set, wherein the decision information system S=<U,C∪D,V,f>Wherein u= { x 1 ,…,x i ,…,x N Sample set of CT image data of pneumonia, N represents the number of CT images of pneumonia, x i Represents the ith sample, x N Representing the nth sample; c= { a 1 ,…,a i ,…,a n Watch (S) } tableA non-empty finite set showing the properties of the CT image data of pneumonia, n represents the number of properties in the CT image data of pneumonia, a i Represents the ith attribute, a n Represents an nth attribute; d= { D 1 ,…,d i ,…,d m The non-empty finite set of decision attributes of the CT image data of the pneumonia is represented, m represents the number of decision categories in the CT image data of the pneumonia, and d i Represents the i-th decision category, d m Representing an mth decision category; v= U-shaped gate a∈C∪D V a ，V a Is the possible case of the CT image data attribute a of pneumonia; u x (C U D) V is an information function which assigns an information value to each CT image, i.ex∈U,f(x,a)∈V a ；
S12, dividing the pneumonia CT image data set S into t data subsets according to the number of different information values of the decision attribute D in the data set, and meeting the requirements wherein Si Represents the ith subset of data, S j Represents the j-th subset of data, i+.j, i=1, 2, …, t, j=1, 2, …, t.
The conversion of the original data set into a four-tuple decision information system is as follows:
wherein u= { x 1 ,x 2 ,…,x 10 The total of 10 samples in this embodiment, c= { a 1 ,a 2 ,…,a 5 Data attributes of the pneumonia CT image are expressed as respiratory rate, white blood cells, platelets, blood pressure and age, respectively, and normalized.
The step S20 includes the steps of:
s21, setting a pneumonia CT image data set as X, setting a reconstruction weight matrix of the X as W, and calculating to-be-examined through a sparse constraint functionSample x of inspection i And other samples { x } 1 ,x 2 ,…,x i-1 ,x i+1 ,…,x N Sparse correlation between samples x to be examined i The objective function of (2) is as follows:
wherein ,xi-1 Represents the i-1 th sample, x i+1 Representing the (i + 1) th sample,is the square of the Frobenius norm, F is a selection mode of norms, I.I.I.I 1 Is a 1-norm, T is a transposed transform of the matrix, η and μ are parameters having values between 0 and 1, R (W) =Tr (W T X T LXW) is a locally preserved projection regularization term, where L is a laplace matrix representing the relationship information between attributes, and Tr is the trace of the matrix, i.e., the sum of the diagonal elements of the matrix.
S22, according to a weight matrix W obtained by reconstructing the CT image data set of the pneumonia, sparse correlation among samples can be seen, if the sparse correlation is larger than 0, the two samples are considered to be in positive correlation, the two samples are put into a sparse neighborhood set, and the number of the samples in positive correlation is recorded as k. For the following Sample x i Sparse neighborhood set SN derived from sparse representation A (x i ) The definition is as follows:
SN A (x i )＝{x j |x j ∈U,W(x i ,x j )＞0}
wherein A represents a subset of attributes in a non-null finite set of data attributes C, x j Represents the jth sample, W (x i ,x j ) Represents x i and xj Sparse correlation weight between. According to the sparse representation, a reconstruction weight matrix W can be obtained, which is expressed as a correlation between the CT image samples of pneumonia. For ease of understanding, sample x is referred to herein 1 To make an expanded description, sample x 1 The sparse correlation matrix W of (2) is represented as follows:
W＝[-0.3063 0 0.5823 0 0.1555 -0.0630 -0.6656 0.5211 0.8318]
it can be seen that sample x 1 And sample x 4 ,x 6 ,x 9 and x10 Form a positive correlation relationship with each other, and x is derived from the correlation 1 Is { x } the sparse neighborhood set 1 ,x 4 ,x 6 ,x 9 ,x 10 }。
S23 calculating CT image sample x of pneumonia i And other samples x j Between Euclidean distance dis (x) i ,x j ) It is defined as follows:
wherein f (x, a) t ) Representing sample x at attribute a of t t Is a value of (b). S24, according to Euclidean distance and k value, adopting k neighbor thought, namely selecting the first k neighbors according to the distance, and solving a distance neighborhood set. For the followingSample x i Distance neighborhood set KN from k neighbors A (x i ) The definition is as follows:
wherein ,representing distance x i N-th sample of>Represents the distance x i The distance of the nearest nth sample. From S22, a sample x can be derived 1 Positive correlation is present between 4 samples, so k=4. Sample x 1 The distances from the other 9 samples are expressed as follows:
dis＝[0.6765 0.3453 0.2593 1.3251 1.0378 1.1568 2.0710 2.07142.0648]
so x is derived from k nearest neighbor 1 Is { x } for a distance neighborhood set 1 ,x 2 ,x 3 ,x 4 ,x 6 }。
S25, fusing the sparse neighborhood set and the distance neighborhood set of the CT image data of the pneumonia to obtain a multi-view neighborhood set MN of the CT image data set of the pneumonia A (x i ) The method comprises the following steps:
MN A (x i )＝{x j |x j ∈SN A (x i )∪x j ∈KN A (x i )}
after fusion, x 1 Is { x } for the multi-view neighborhood set 1 ,x 2 ,x 3 ,x 4 ,x 6 ,x 9 ,x 10 }。
S26, performing first-order deletion operation on the fused neighborhood set, and calculating a sample x in the CT (computed tomography) image data set i Sample x in its neighborhood set j The average distance of (2) is denoted ave. Comparing dis (x) i ,x j ) And ave, if dis (x i ,x j )>ave, consider sample x j And sample x i The first order neighborhood relation is not satisfied, and the sample x is obtained j From x i Is deleted from the multi-view neighborhood set. For the followingSample x i The first-order neighborhood set obtained after the first deletion isThe calculation formula is as follows:
respectively calculate x 1 Distance from the samples in the fused neighborhood set and average distance ave= 1.0668 is found. Due to x 1 To x 9 and x10 Since the distance of (a) exceeds ave, it is deleted from the neighborhood set, and the first-order neighborhood set after deletion is { x } 1 ,x 2 ,x 3 ,x 4 ,x 6 }。
And S27, performing second-order operation on a first-order neighborhood set of samples in the CT image data set of the pneumonia according to the midline theorem, and deleting samples which do not meet the second-order requirement. Determining sample x in a neighborhood set using a perpendicular bisector j Whether or not the sample x exists on the perpendicular bisector i Is provided. Count all samples in the neighborhood set if sample x j Sample x to be examined i If the value is not changed, otherwise the value is increased by 1. After counting, consider the sample with the largest value as the sample x to be examined i The two-order neighborhood relation is not satisfied, and the two-order neighborhood relation is regarded as a redundant sample. And deleting the redundant samples to obtain a second-order neighborhood set of the CT image data samples of the pneumonia. According to the theorem of the perpendicular bisectorSample x i The second order neighborhood set obtained after the second deletion is +.>The definition of puncturing is as follows:
wherein count represents the calculation of the sample x from the perpendicular bisector j Whether to be examined sample x i The values on the same side, n', represent the sample x to be examined i Is the number in the first order neighborhood set,representing the calculation of x from the perpendicular bisector i Samples in the first order neighborhood are not in sample x to be examined i Maximum on the same side. According to the midline theorem, it is obtained through analysis that for sample x 1 Its neighborhood x 2 ,x 3 ,x 4 ,x 6 The obtained count values are 3,2,2,3, respectively, so the maximum count value is 3, and the corresponding sample is x 2 and x6 It is deleted from the neighborhood set. Finally, the second order neighborhood set after the secondary deletion is { x } 1 ,x 3 ,x 4 }。
The step S30 includes the steps of: s31 calculating the p-th sample x to be examined in the CT image data set of the pneumonia p And other sample x of q q Distance dis (x) p ,x q ) And sparse representation W (x p ,x q ) Respectively denoted as dis p,q and spap,q According to dis p,q and spap,q Calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)
S32, calculating other samples x in the CT image dataset of the pneumonia q For the sample x to be examined p Evidence information provided, Ω=u/d= { D 1 ,D 2 ,...,D j The decision class division set of the CT image dataset of the pneumonia is shown as the figure of j, D j For the j decision class division, if the CT image sample x of the pneumonia q Class label of l t And decision class division into D t Then (x) q ,D t ) Can be independently the sample x to be examined p Providing evidence support, including evidence informationThe definition is as follows:
wherein ,Dt For the t decision class division, w 1 and w2 Importance weight duty cycle representing distance measure and sparse representation, w 1 +w 2 =1. μ is a correlation parameter of the distance metric, η is a correlation parameter of the sparse representation, and μ > 0 and η > 0 are satisfied.
In this embodiment, w is taken 1 ＝0.4，w 2 =0.6, can calculate the CT image sample x to be inspected 1 And neighborhood sample x 3 Evidence information between
S33 calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)The middle class label is l t Is as followsSample set->For x p Evidence support of (a) can be expressed as +.>Namely:
calculating CT image sample x to be inspected 1 Is a second order neighborhood of (2)The middle class label is l 1 Is a sample set of (1)The sample set->For x 1 Evidence support of->
S34 calculating CT image sample x of pneumonia p Evidence information under different category labelsAnd performs a merging operation on the certification information, which may be specifically expressed as:
wherein ,j is the number of decision divisions. Calculating CT image sample x to be inspected 1 Evidence information under different category tags +.>
S35 calculating CT image sample x of pneumonia p Dividing decision classes Ω=u/d= { D 1 ,D 2 ,...,D j D in } t Trust function for classThe specific definition is as follows:
calculating CT image sample x to be inspected 1 For D 1 Trust function of (c)
S36 calculating CT image sample x of pneumonia p Neighborhood evidence entropy H (d|a), x p Is set as the second order neighborhood of (2)[x p ]D is x p Is defined as follows:
wherein, |U| is the number of domains.
The step S40 includes the steps of: s41 calculating each pathological attribute in the CT image dataset of the pneumoniaExternal importance measure sig with respect to reduced set red out (red, a, D) =h (d|red) -H (d|red u { a }) and selects the attribute a corresponding to the greatest attribute importance k ；
S42 judging sig out (red,a k D) if it is greater than 0, if sig is satisfied out (red,a k D) > 0, attribute a will be k Adding the candidate attribute subset C into the reduced set red, and deleting the candidate attribute subset C;
s43 calculates and compares H (D|red) and H (D|C), if H (D|red) > H (D|C) and sig out (red,a k D) is less than or equal to 0, attribute a k Delete from reduced set red until
S44 for each attribute in the reduced set redRe-calculating the importance measure, sig, of the attribute in (red,a w ,D)＝H(D|red-{a w -H (d|red), deleting attribute a in turn w Comparison of sig in (red,a w D) if it is greater than 0, if sig in (red,a w D) is less than or equal to 0, attribute a w Deleted from the reduced set red. And (3) after sequentially circularly traversing all the attributes in red, outputting a reduced set red of the pneumonia CT image data set.
The foregoing description is only exemplary embodiments of the present invention and is not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes using the descriptions and the drawings of the present invention or directly or indirectly applied to other related technical fields are included in the scope of the present invention.

Claims (5)

1. A method for reducing the attributes of a CT image of pneumonia based on multi-view neighborhood evidence entropy is characterized by comprising the following steps:
s10, reading a pneumonia CT image data set, and converting the data set into a four-tuple decision information system;
s20, representing the correlation between the CT image data samples of the pneumonia through two views of sparse representation and distance measurement, and describing a neighborhood range through two stages of fusion and deletion;
s30, fusing the evidence theory and the neighborhood entropy to construct a neighborhood evidence entropy for describing the attribute importance of the CT image data of the pneumonia;
s40, calculating an attribute reduction set in the CT image dataset of the pneumonia.
2. The method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy according to claim 1, wherein said step S10 comprises the steps of:
s11, reading a pneumonia CT image data set, determining an attribute set and decision classification of the pneumonia CT image data set, wherein the decision information system S=<U,C∪D,V,f>Wherein u= { x 1 ,…,x i ,…,x N Sample set of CT image data of pneumonia, N represents the number of CT images of pneumonia, x i Represents the ith sample, x N Representing the nth sample; c= { a 1 ,…,a i ,…,a n The non-empty finite set of the attribute of the CT image data of the pneumonia is represented, n represents the number of the attribute in the CT image data of the pneumonia, a i Represents the ith attribute, a n Represents an nth attribute; d= { D 1 ,…,d i ,…,d m The non-empty finite set of decision attributes of the CT image data of the pneumonia is represented, m represents the number of decision categories in the CT image data of the pneumonia, and d i Represents the i-th decision category, d m Representing an mth decision category; v= U-shaped gate a∈C∪D V a ，V a Is the possible case of the CT image data attribute a of pneumonia; u x (C U D) V is an information function which assigns an information value to each CT image, i.e
S12, dividing the pneumonia CT image data set S into t data subsets according to the number of different information values of the decision attribute D in the data set, and meeting the requirements of wherein Si Represents the ith subset of data, S j Represents the j-th subset of data, i+.j, i=1, 2, …, t, j=1, 2, …, t.
3. The method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy according to claim 1, wherein said step S20 comprises the steps of:
s21, setting a pneumonia CT image data set as X, setting a reconstruction weight matrix of X as W, and calculating a sample X to be inspected through a sparse constraint function i And other samples { x } 1 ,x 2 ,…,x i-1 ,x i+1 ,…,x N Sparse correlation between samples x to be examined i The objective function of (2) is as follows:
wherein ,xi-1 Represents the i-1 th sample, x i+1 Representing the (i + 1) th sample,is the square of the Frobenius norm, F is a selection mode of norms, I.I.I.I 1 Is a 1-norm, T is a transposed transform of the matrix, η and μ are parameters having values between 0 and 1, R (W) =Tr (W T X T LXW) is a locally preserved projection regularization term, where L is a laplace matrix representing relationship information between attributes, tr is the trace of the matrix, i.e., the sum of diagonal elements of the matrix;
s22, according to a weight matrix W obtained by reconstructing a pneumonia CT image data set, finding out sparse correlation among samples, if the sparse correlation is larger than 0, considering that the two samples are in positive correlation, putting the two samples into a sparse neighborhood set, and recording the number of samples in positive correlation as k, wherein for the samples with positive correlationSample x i Sparse neighborhood set SN derived from sparse representation A (x i ) The definition is as follows:
SN A (x i )＝{x j |x j ∈U,W(x i ,x j )＞0}
wherein A represents a subset of attributes in a non-null finite set of data attributes C, x j Represents the jth sample, W (x i ,x j ) Represents x i and xj Sparse correlation weight values between;
s23, calculating CT image sample x of pneumonia i And other samples x j Between Euclidean distance dis (x) i ,x j ) It is defined as follows:
wherein f (x, a) t ) Representing sample x at attribute a of t t Is a value of (b);
s24, according to Euclidean distance and k value, adopting k neighbor thought, namely selecting the first k neighbors according to distance, solving a distance neighbor set, and for the distance Sample x i Distance neighborhood set KN from k neighbors A (x i ) The definition is as follows:
wherein ,representing distance x i N-th sample of>Represents the distance x i The distance of the nearest nth sample;
s25, fusing the sparse neighborhood set and the distance neighborhood set of the CT image data of the pneumonia to obtain a multi-view neighborhood set MN of the CT image data set of the pneumonia A (x i ) The method comprises the following steps:
MN A (x i )＝{x j |x j ∈SN A (x i )∪x j ∈KN A (x i )}
s26, performing first-order deletion operation on the fused neighborhood set, and calculating a pneumonia CT image data setIs of sample x of (2) i Sample x in its neighborhood set j Is denoted as ave, compares dis (x i ,x j ) And ave, if dis (x i ,x j )>ave, consider sample x j And sample x i The first order neighborhood relation is not satisfied, and the sample x is obtained j From x i For deletion in a multi-view neighborhood set of (2)Sample x i The first-order neighborhood set obtained after the first deletion isThe calculation formula is as follows:
s27, performing second-order operation on a first-order neighborhood set of samples in the CT image data set of the pneumonia according to the midline theorem, deleting samples which do not meet second-order requirements, and determining a sample x in the neighborhood set by using a perpendicular bisector j Whether or not the sample x exists on the perpendicular bisector i Counting all samples in the neighborhood set if sample x j Sample x to be examined i If the numerical value is not changed, otherwise, the numerical value is increased by 1, and if the counting is finished, the sample with the largest numerical value is considered to be the sample x to be examined i The two-order neighborhood relation is not satisfied, the two-order neighborhood relation is regarded as a redundant sample, the redundant sample is deleted to obtain a second-order neighborhood set of the CT image data sample of the pneumonia, and according to the midplumb line theorem, the two-order neighborhood set of the CT image data sample of the pneumonia is regarded as a redundant sampleSample x i The second order neighborhood set obtained after the second deletion is +.>The definition of puncturing is as follows:
wherein count represents the calculation of the sample x from the perpendicular bisector j Whether to be examined sample x i The values on the same side, n', represent the sample x to be examined i Is the number in the first order neighborhood set,representing the calculation of x from the perpendicular bisector i Samples in the first order neighborhood are not in sample x to be examined i Maximum on the same side.
4. The method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy according to claim 1, wherein said step S30 comprises the steps of:
s31, calculating a p-th sample x to be examined in a pneumonia CT image data set p And other sample x of q q Distance dis (x) p ,x q ) And sparse representation W (x p ,x q ) Respectively denoted as dis p,q and spap,q According to dis p,q and spap,q Calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)
S32, calculating other samples x in the CT image dataset of the pneumonia q For the sample x to be examined p Evidence information provided, Ω=u/d= { D 1 ,D 2 ,...,D j The decision class division set of the CT image dataset of the pneumonia is shown as the figure of j, D j For the j decision class division, if the CT image sample x of the pneumonia q Class label of l t And decision class division into D t Then (x) q ,D t ) Independently be the sample x to be examined p Providing evidence support, including evidence informationThe definition is as follows:
wherein ,Dt For the t decision class division, w 1 and w2 Importance weight duty cycle representing distance measure and sparse representation, w 1 +w 2 =1, μ is a correlation parameter of the distance metric, η is a correlation parameter of the sparse representation, and μ > 0 and η > 0 are satisfied;
s33, calculating CT image sample x of pneumonia p Is a second order neighborhood of (2)The middle class label is l t Is as followsSample set->For x p Evidence support of (1) is expressed as->Namely:
s34, calculating CT image sample x of pneumonia p Evidence information under different category labelsAnd performs a merging operation on the certification information, which is specifically expressed as:
wherein ,j is the number of decision divisions;
s35, calculating CT image sample x of pneumonia p Dividing decision classes Ω=u/d= { D 1 ,D 2 ,...,D j D in } t Trust function for classThe specific definition is as follows:
s36, calculating CT image sample x of pneumonia p Neighborhood evidence entropy H (d|a), x p Is set as the second order neighborhood of (2)[x p ] D Is x p Is defined as follows:
wherein, |U| is the number of domains.
5. The method for reducing the attributes of a pneumonia CT image based on multi-view neighborhood evidence entropy according to claim 1, wherein said step S40 comprises the steps of:
s41, calculating each pathological attribute in the CT image dataset of the pneumoniaExternal importance measure sig with respect to reduced set red out (red, a, D) =h (d|red) -H (d|red u { a }) and selects the attribute a corresponding to the greatest attribute importance k ；
S42, judging sig out (red,a k D) if it is greater than 0, if sig is satisfied out (red,a k D) > 0, attribute a will be k Adding the candidate attribute subset C into the reduced set red, and deleting the candidate attribute subset C;
s43, calculate and compare H (D|red) and H (D|C), if H (D|red) > H (D|C) and sig out (red,a k D) is less than or equal to 0, attribute a k Delete from reduced set red until
S44, for each attribute in the reduced set redRe-calculating the importance measure, sig, of the attribute in (red,a w ,D)＝H(D|red-{a w -H (d|red), deleting attribute a in turn w Comparison of sig in (red,a w D) if it is greater than 0, if sig in (red,a w D) is less than or equal to 0, attribute a w Deleted from reduced set red, all attributes in red are cycled through in turnAfter that, a reduced set red of the pneumonia CT image dataset is output.
CN202310226858.9A 2023-03-07 2023-03-07 Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy Pending CN116596836A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310226858.9A CN116596836A (en) 2023-03-07 2023-03-07 Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310226858.9A CN116596836A (en) 2023-03-07 2023-03-07 Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy

Publications (1)

Publication Number Publication Date
CN116596836A true CN116596836A (en) 2023-08-15

Family

ID=87603287

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310226858.9A Pending CN116596836A (en) 2023-03-07 2023-03-07 Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy

Country Status (1)

Country Link
CN (1) CN116596836A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117542503A (en) * 2023-10-16 2024-02-09 南通大学 Schizophrenia multi-view classification method based on dynamic evidence fusion neural network

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117542503A (en) * 2023-10-16 2024-02-09 南通大学 Schizophrenia multi-view classification method based on dynamic evidence fusion neural network

Similar Documents

Publication Publication Date Title
CN114926746B (en) SAR image change detection method based on multiscale differential feature attention mechanism
CN113314205B (en) Efficient medical image labeling and learning system
He et al. Automated model design and benchmarking of deep learning models for covid-19 detection with chest ct scans
CN113256636B (en) Bottom-up parasite species development stage and image pixel classification method
CN108305253A (en) A kind of pathology full slice diagnostic method based on more multiplying power deep learnings
CN115359873B (en) Control method for operation quality
CN114549469A (en) Deep neural network medical image diagnosis method based on confidence degree calibration
CN114821299B (en) Remote sensing image change detection method
CN116596836A (en) Pneumonia CT image attribute reduction method based on multi-view neighborhood evidence entropy
Liu et al. BrainTGL: A dynamic graph representation learning model for brain network analysis
CN116029953A (en) Non-reference image quality evaluation method based on self-supervision learning and transducer
CN109935321B (en) Risk prediction system for converting depression patient into bipolar affective disorder based on functional nuclear magnetic resonance image data
Zhou et al. Adaptive weighted locality-constrained sparse coding for glaucoma diagnosis
CN105426836A (en) Single-sample face recognition method based on segmented model and sparse component analysis
Sun et al. Boosting robust learning via leveraging reusable samples in noisy web data
Lonij et al. Open-world visual recognition using knowledge graphs
CN110929877B (en) Model building method, device, equipment and storage medium based on transfer learning
Shi et al. Domain adaptation based on rough adjoint inconsistency and optimal transport for identifying autistic patients
Wang et al. A comprehensive survey on deep active learning and its applications in medical image analysis
Cao et al. No-reference image quality assessment by using convolutional neural networks via object detection
Biesner et al. Improving chest x-ray classification by RNN-based patient monitoring
CN116070120B (en) Automatic identification method and system for multi-tag time sequence electrophysiological signals
CN117974693B (en) Image segmentation method, device, computer equipment and storage medium
Mohammed et al. Graph Neural Networks for Anomaly Detection in Internet of Medical Things
Meng¹ et al. Check for updates

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