CN113920071A - New coronavirus image identification method based on convolutional neural network algorithm - Google Patents
New coronavirus image identification method based on convolutional neural network algorithm Download PDFInfo
- Publication number
- CN113920071A CN113920071A CN202111134598.XA CN202111134598A CN113920071A CN 113920071 A CN113920071 A CN 113920071A CN 202111134598 A CN202111134598 A CN 202111134598A CN 113920071 A CN113920071 A CN 113920071A
- Authority
- CN
- China
- Prior art keywords
- image
- covid
- layer
- algorithm
- neural network
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 48
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 35
- 241000711573 Coronaviridae Species 0.000 title claims abstract description 15
- 208000025721 COVID-19 Diseases 0.000 claims abstract description 58
- 238000013135 deep learning Methods 0.000 claims abstract description 16
- 238000001514 detection method Methods 0.000 claims abstract description 14
- 238000012706 support-vector machine Methods 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 12
- 201000003176 Severe Acute Respiratory Syndrome Diseases 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 9
- 238000007619 statistical method Methods 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 238000007635 classification algorithm Methods 0.000 claims description 5
- 210000004027 cell Anatomy 0.000 claims description 3
- 230000006835 compression Effects 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims 1
- 238000005070 sampling Methods 0.000 claims 1
- 230000011218 segmentation Effects 0.000 abstract description 7
- 238000012545 processing Methods 0.000 abstract description 2
- 201000010099 disease Diseases 0.000 description 7
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 7
- 208000024891 symptom Diseases 0.000 description 5
- 208000015181 infectious disease Diseases 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 230000000877 morphologic effect Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 208000001528 Coronaviridae Infections Diseases 0.000 description 2
- 208000025370 Middle East respiratory syndrome Diseases 0.000 description 2
- 206010035664 Pneumonia Diseases 0.000 description 2
- 241000700605 Viruses Species 0.000 description 2
- 210000004072 lung Anatomy 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 206010011224 Cough Diseases 0.000 description 1
- 206010012735 Diarrhoea Diseases 0.000 description 1
- 206010025080 Lung consolidation Diseases 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 206010028735 Nasal congestion Diseases 0.000 description 1
- 206010068319 Oropharyngeal pain Diseases 0.000 description 1
- 201000007100 Pharyngitis Diseases 0.000 description 1
- 206010037660 Pyrexia Diseases 0.000 description 1
- 206010057190 Respiratory tract infections Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 210000003792 cranial nerve Anatomy 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 208000017574 dry cough Diseases 0.000 description 1
- 206010016256 fatigue Diseases 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 238000007620 mathematical function Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 201000009240 nasopharyngitis Diseases 0.000 description 1
- 238000003752 polymerase chain reaction Methods 0.000 description 1
- 238000002600 positron emission tomography Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 206010040882 skin lesion Diseases 0.000 description 1
- 231100000444 skin lesion Toxicity 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/45—Analysis of texture based on statistical description of texture using co-occurrence matrix computation
-
- 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
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- 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
- G16H50/80—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for detecting, monitoring or modelling epidemics or pandemics, e.g. flu
-
- 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
- G06T2207/10056—Microscopic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Primary Health Care (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a new coronavirus image identification method based on a convolutional neural network algorithm, which utilizes three different classifiers: the method comprises the steps that an SVM, a KNN and a CNN are used for simulating and evaluating the proposed COVID-19 detection model, images of COVID-19, SARS and normal persons are collected to verify the model, firstly, effective methods such as preprocessing, region growing segmentation and deep learning classification are used, a COVID-19 early prediction method based on processing is provided, and a mixed feature extraction method of DWT, geometric features and texture features is adopted after segmentation. Then, classifying by adopting a support vector machine and a KNN learning algorithm; in a second scenario, where the CNN algorithm is applied to COVID-19 classification, the method will help medical systems to quickly improve the detection efficiency of COVID-19.
Description
Technical Field
The invention relates to the field of deep learning and new coronavirus image identification, in particular to a new coronavirus image identification method based on a convolutional neural network algorithm.
Background
Coronaviruses comprise a large number of virus species that can cause disease in animals and humans, many of which cause respiratory infections in humans, ranging from the common cold to more severe diseases such as Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS).
The most common symptoms of covi-19 are fever, fatigue and dry cough, some patients may also experience pain, nasal congestion, chills, sore throat and/or diarrhea, some patients may be infected with the virus without any symptoms, the symptoms of covi-19 are usually mild and gradual, most people (80%) may recover from the disease without special treatment, and about one-sixth of those infected with covi-19 may develop severe symptoms.
Chest image screening has become a common diagnostic tool for pneumonia, and in addition, images play an important role in the quantitative assessment of covi-19 and disease monitoring, on which the areas of covi-19 infection at the initial stage of infection can be distinguished by the Glass nodules (GGO) in the lungs, and the areas at the later stage of infection can be distinguished by lung consolidation, compared to reversing the polymerase chain reaction, several studies have shown that images are more sensitive and effective for covi-19 screening, and chest image imaging is more sensitive to covi-1 detection even without clinical symptoms, and physicians typically diagnose whether a patient's lungs are infected with new crown pneumonia by viewing the patient's image.
Many machine learning algorithms have been used to enhance medical system identification of disease, and biomedical images from different devices, such as X-ray, computed tomography (images), Magnetic Resonance Imaging (MRI), and Positron Emission Tomography (PET), provide important information to distinguish normal and abnormal patients, extract this information from the images, and then identify disease and infection using machine learning algorithms.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a new coronavirus image identification method based on a convolution neural network algorithm.
The technical scheme adopted by the invention is that three different classifiers are utilized: support Vector Machines (SVMs), k-nearest neighbors (k-NN) and deep learning Convolutional Neural Network (CNN) algorithms to simulate and evaluate the proposed codv-19 detection model, which is validated by collecting images of codv-19, SARS and normal persons, using both conditions and evaluation metrics to complete the method.
In the present invention, a standard data set of patients diagnosed with COVID-19, patients diagnosed with SARS and normal patients is collected, and the method is performed using two scenarios, as follows:
and 3, evaluating the COVID-19 classification and detection model by adopting the evaluation index.
In the invention, in step 1, the Gaussian filtering technology is adopted to remove the distortion and the blur of the image and improve the image quality, and the Gaussian function is calculated by using a formula (1) as follows:
where x is the distance from the origin on the horizontal axis, y is the distance from the origin on the vertical axis, and σ is the standard deviation of the Gaussian distribution.
In the invention, in step 2, texture features are extracted from the COVID-19 image by using a gray level co-occurrence matrix method, and important features are obtained by adopting five statistical analysis methods of mean value, energy, contrast, uniformity and correlation, wherein the five statistical analysis methods are shown in the following formula:
the mean value is used to measure the average intensity of the COVID-19 image pixels,
the histogram of the COVID-19 image pixels is normalized by energy,
contrast is used to measure the local intensity variance of COVID-19 image pixels,
uniformity to measure uniformity of COVID-19 image pixels,
the correlation is used to measure the gray scale correlation distribution,
where x and y are the locations of the pixels in the cell, m and n are the column and row numbers, respectively, μ is the mean number, and σ is the standard deviation.
In the invention, step 2, the gradient is a vector consisting of size and direction, the gradient is calculated by adopting a gradient method, the derivative of the gradient is horizontal and vertical, a Prewitt operator is used for detecting the edge and extracting the vector value, and the operator detects the edge by utilizing the external value of the gray level change of each point around the image pixel;
p is the original image, GxAnd GyIs the value of the image gradient, G, detected by the horizontal and vertical edgesxAnd GyIs calculated as follows:
in the present invention, in step 2, the feature extraction using the discrete wavelet transform method, a wavelet decomposition process, in which 4 subband images (HH, HL, LL, and LH) representing data of each size, the LL subband being an object of extracting features of an image, the subband being an estimation component, and HH, H, and HL taking into account specific elements of the image, features are extracted from the image using the 3-level decomposition of the Harr wavelet.
In the invention, step 2, the model for detecting COVID-19 by three classification methods of COVID-19, SARS and normal people is completed by using SVM algorithm, as shown in the following formula,
In the invention, step 2. non-parameter classification is completed by using a K-NN algorithm, and the algorithm searches the nearest point between the features by using an Euclidean distance method, as follows:
wherein x1-x2, y1-y2 represent Euclidean vectors.
In the invention, step 2, a deep learning algorithm (CNN) is adopted to classify the COVID-19, wherein the CNN consists of four layers, namely an input layer, a convolution layer, a polling layer and a connecting layer, and is used for extracting features from the COVID-19 image, and the operation steps of the CNN are as follows:
an input layer: inputting a COVID-19 image, extracting dataset image tags, which also standardizes data scaling to speed up the training dataset, this layer being used as a pre-processing step to improve accuracy;
and (3) rolling layers: the characteristic non-linearity of convolutional layers is mapped to multiple network layers to obtain better results, this layer uses an activation function called a corrected linear unit (ReLU) which is applied to the output of the previous layer, the ReLU function being defined as the positive part of its parameters, as follows:
f(x)=x+=max(0,x);
and a polling layer: this layer improves the training data by reducing the dimensionality and compression of the data, thereby improving the accuracy of the classification, and performs downsampling along the spatial dimension of a given input, thereby reducing the number of parameters in the activation;
connecting layers: this layer calculates the output by computing the weights of the neurons locally associated with the input and the dot product connecting them with a small region of the input volume, the ReLU being applied to the output of the previous layer activation.
In the invention, step 3. the statistical analysis index is used for evaluating the proposed COVID-19 classification and detection model, and the evaluation index (namely, accuracy, specificity, sensitivity and recall rate) is adopted and calculated as follows:
the detected image can be finally determined as TP, TN, FP and FN, wherein TP is true positive, TN is true negative, FP is false positive and FN is false negative.
Drawings
FIG. 1 is a flow chart framework of the present invention
FIG. 2 is a sample of a standard data set of the present invention
FIG. 3 is the Prewitt operator mask of the present invention
FIG. 4 is a flowchart of the wavelet decomposition procedure of the present invention
FIG. 5 is a block diagram of a convolutional neural network of the present invention
Detailed Description
It should be noted that the embodiments and features of the embodiments can be combined with each other without conflict, and the present application will be further described in detail with reference to the drawings and specific embodiments.
As shown in fig. 1, a new coronavirus image identification method based on convolutional neural network algorithm utilizes three different classifiers: support Vector Machine (SVM), k-nearest neighbor (k-NN), and deep learning Convolutional Neural Network (CNN) algorithms to simulate and evaluate the proposed COVID-19 detection model, which is validated by collecting images of COVID-19, SARS, and normal humans, using both conditions and evaluation metrics to complete the method.
In the present invention, a standard data set of patients diagnosed with COVID-19, patients diagnosed with SARS and normal patients is collected, and the method is performed using two scenarios, as follows:
and 3, evaluating the COVID-19 classification and detection model by adopting the evaluation index.
The images need to be pre-processed before being classified by the algorithm, and the main purpose of the pre-processing stage is to suppress any distortion in the images, thereby improving the ability to obtain important features. Furthermore, the pre-processing step is very important to improve low contrast and high noise levels. A detailed description of these pre-treatment stages is given below
In the invention, in step 1, the Gaussian filtering technology is adopted to remove the distortion and the blur of the image and improve the image quality, and the Gaussian function is calculated by using a formula (1) as follows:
where x is the distance from the origin on the horizontal axis, y is the distance from the origin on the vertical axis, and σ is the standard deviation of the Gaussian distribution.
In order to improve the segmentation quality, a morphological segmentation method is adopted. When the COVID-19 image is segmented, small spots are left that are not skin lesions. These small spots must be removed in order to obtain the proper information from the image. Morphological segmentation, or over-segmentation, avoids confusion between isolated artifacts and objects of interest by preventing the detection of very small non-lesions. Thus, morphological segmentation is used to remove very small objects from the binary COVID-19 image, while preserving the shape and size of larger objects. Fig. 3 shows the various stages of the segmentation method by which the morphologic method extracts and isolates the shapes with the largest area.
The gray level co-occurrence matrix method is a feature extraction method for extracting important features from a biomedical image by utilizing statistical analysis, and the method is used for extracting texture features from a COVID-19 image.
the mean value is used to measure the average intensity of the COVID-19 image pixels,
the histogram of the COVID-19 image pixels is normalized by energy,
contrast is used to measure the local intensity variance of COVID-19 image pixels,
uniformity to measure uniformity of COVID-19 image pixels,
the correlation is used to measure the gray scale correlation distribution,
where x and y are the locations of the pixels in the cell, m and n are the column and row numbers, respectively, μ is the mean, and σ is the standard deviation.
In the invention, step 2. the gradient is a vector composed of size and direction, the gradient is calculated by a gradient method, the derivative of the gradient is horizontal and vertical, a Prewitt operator is used for detecting edges, and vector values are extracted, as shown in figure 4. The operator detects edges using an external value of the gray scale variation of each point around the image pixel;
p is the original image, GxAnd GyIs the value of the image gradient, G, detected by the horizontal and vertical edgesxAnd GyIs calculated as follows:
in the present invention, step 2. the Discrete Wavelet Transform (DWT) method is more efficient than the wavelet transform in consideration of the dimension and position of the binary. Fig. 5 shows a wavelet decomposition process in which 4 subband images (HH, HL, LL, and LH) represent data of each size. In the invention, the LL subband is the target for extracting image features; the subband is an estimation component, and HH, H, and HL consider a specific element of the image. Features are extracted from the image using a 3-level decomposition of the Harr wavelet.
In the invention, the support vector machine algorithm in the step 2 is a strong and large machine learning algorithm for classification and regression, the support vector machine algorithm is used for dichotomy classification, such as two-class binary classification or multi-class binary classification, and the algorithm is applied to a training data set to obtain a classification model allowing an optimal decision function; it maps non-linear separable data from a low-dimensional space to a high-dimensional space; the SVM algorithm classifies data by finding a hyperplane that distinguishes two classes of data.
When a larger margin is set, the algorithm of the support vector machine can obtain a lower error rate; the invention adopts three classification methods of COVID-19, SARS and normal person to establish a model for detecting COVID-19; wherein the radial basis functions and the kernel functions have a higher accuracy. As shown in the following formula,
In the invention, step 2, non-parameter classification is completed by using a K-NN algorithm, wherein the K-NN is a non-parameter classification algorithm, and has lower complexity compared with other classification algorithms, so the algorithm is called as an inert algorithm. The K-NN algorithm classifies data using nearest neighbor features. The algorithm uses the euclidean distance method to find the closest points between features as follows:
wherein x1-x2, y1-y2 represent Euclidean vectors.
In the invention, step 2, a deep learning algorithm (CNN) is adopted to classify the COVID-19 as a part of machine learning, and the deep learning relates to a technology for simulating a human cranial nerve system. Using algorithms, deep learning simulation reasoning and data extraction, which contains hidden layers of mathematical functions for analyzing specific patterns in specific data; deep learning becomes increasingly important in the classification and detection of images and video, and has many uses in both the medical field (for X-ray identification) and in computer vision applications. The application of deep learning in the medical field improves image quality and prediction of different types of diseases. Convolutional Neural Network (CNN) is a neural network, the most common deep learning algorithm; these are feed-forward neural networks, CNN consists of four layers, namely an input layer, a convolutional layer, a polling layer and a connection layer, for extracting features from COVID-19 images, and the operation of CNN is as follows:
an input layer: inputting a COVID-19 image, extracting dataset image tags, which also standardizes data scaling to speed up the training dataset, this layer being used as a pre-processing step to improve accuracy;
and (3) rolling layers: the characteristic non-linearity of convolutional layers is mapped to multiple network layers to obtain better results, this layer uses an activation function called a corrected linear unit (ReLU) which is applied to the output of the previous layer, the ReLU function being defined as the positive part of its parameters, as follows:
f(x)=x+=max(0,x);
and a polling layer: this layer improves the training data by reducing the dimensionality and compression of the data, thereby improving the accuracy of the classification, and performs downsampling along the spatial dimension of a given input, thereby reducing the number of parameters in the activation;
connecting layers: this layer calculates the output by computing the weights of the neurons locally associated with the input and the dot product connecting them with a small region of the input volume, the ReLU being applied to the output of the previous layer activation.
In the invention, step 3. the statistical analysis index is used for evaluating the proposed COVID-19 classification and detection model, and the evaluation index (namely, accuracy, specificity, sensitivity and recall rate) is adopted and calculated as follows:
the detected image can be finally determined as TP, TN, FP and FN, wherein TP is true positive, TN is true negative, FP is false positive and FN is false negative.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made herein without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims (10)
1. A new coronavirus image identification method based on a convolutional neural network algorithm is characterized in that three different classifiers are utilized: support Vector Machines (SVMs), k-nearest neighbors (k-NN) and deep learning Convolutional Neural Network (CNN) algorithms to simulate and evaluate the proposed codv-19 detection model, which is validated by collecting images of codv-19, SARS and normal humans.
2. The method of claim 1, wherein standard data sets of patients diagnosed with COVID-19, patients diagnosed with SARS and normal patients are collected, and two scenarios are used to perform the method, comprising the following steps:
step 1, denoising a COVID-19 image by adopting a Gaussian filter;
step 2, after denoising treatment, dividing the image into two conditions, wherein in the first condition, a region of interest (ROI) is segmented from the COVID-19 image by using a region growing algorithm, then a mixed feature extraction method combining a gray level co-occurrence matrix (GLCM) method, a gradient method and a Discrete Wavelet Transform (DWT) method is adopted to extract significant features, and the features are processed by a classification algorithm (SVM and k-NN); in the second case, the COVID-19 is classified by using a deep learning algorithm (CNN);
and 3, evaluating the COVID-19 classification and detection model by adopting the evaluation index.
3. The method for identifying the new coronavirus image based on the convolutional neural network algorithm as claimed in claim 2, wherein in step 1, a gaussian filtering technique is adopted to remove distortion and blur of the image and improve the image quality, and a gaussian function is calculated by using a formula (1) as follows:
where x is the distance from the origin on the horizontal axis, y is the distance from the origin on the vertical axis, and σ is the standard deviation of the Gaussian distribution.
4. The method for identifying the new coronavirus image based on the convolutional neural network algorithm as claimed in claim 2, wherein in step 2, a gray level co-occurrence matrix method is used for extracting texture features from the COVID-19 image, and five statistical analysis methods of mean value, energy, contrast, uniformity and correlation are adopted to obtain important features, wherein the following formula is shown as follows:
the mean value is used to measure the average intensity of the COVID-19 image pixels,
the histogram of the COVID-19 image pixels is normalized by energy,
contrast is used to measure the local intensity variance of COVID-19 image pixels,
uniformity to measure uniformity of COVID-19 image pixels,
the correlation is used to measure the gray scale correlation distribution,
where x and y are the positions of the pixels in the cell, m and n are the column and row numbers, respectively, μ is the mean, and σ is the standard deviation.
5. The method of claim 4, wherein the gradient is a vector consisting of magnitude and direction, the gradient is calculated by gradient method, the derivative of which is horizontal and vertical, Prewitt operator is used to detect the edge, and the vector value is extracted, the operator uses the external value of the gray level change of each point around the image pixel to detect the edge;
p is the original image, GxAnd GyIs the value of the image gradient, G, detected by the horizontal and vertical edgesxAnd GyIs calculated as follows:
6. a new coronavirus image recognition method based on convolutional neural network algorithm as set forth in claim 5, characterized in that, in step 2, the feature extraction by discrete wavelet transform method, wavelet decomposition process, in which 4 subband images (HH, HL, LL, and LH) represent data of each size, LL subband is an object of extracting the feature of the image, the subband is an estimation component, and HH, H, and HL extract the feature from the image by 3-level decomposition of Harr wavelet considering the specific elements of the image.
7. The method for recognizing a new coronavirus image based on convolutional neural network algorithm as set forth in claim 6, wherein the step 2. the model for detecting COVID-19 by three classification methods of COVID-19, SARS and normal person is performed by using SVM algorithm, as shown in the following formula,
9. The method for identifying new coronavirus image based on convolutional neural network algorithm as claimed in claim 8, wherein in step 2, the COVID-19 is classified by using deep learning algorithm (CNN), the CNN is composed of four layers, i.e. input layer, convolutional layer, polling layer and connection layer, for extracting features from the COVID-19 image, and the CNN is operated as follows:
input layer, inputting COVID-19 image, extracting data set image label, it also standardizes data scaling to speed up training data set, this layer is used as preprocessing step to improve precision;
convolutional layer-mapping the characteristic non-linearity of a convolutional layer to a multi-layer network layer to obtain better results, this layer uses an activation function called a corrected linear unit (ReLU) which is applied to the output of the previous layer, the ReLU function being defined as the positive part of its parameters, as follows:
f(x)=x+=max(0,x);
a polling layer that improves training data by reducing the dimensionality and compression of the data to improve classification accuracy, and that performs down-sampling along the spatial dimension of a given input to reduce the number of parameters in activation;
connected layer this layer calculates the output by computing the weights of the neurons locally related to the input and the dot product connecting them with a small region of the input volume, the ReLU being applied to the output of the previous layer activation.
10. The method for identifying new coronavirus image based on convolutional neural network algorithm as claimed in claim 2, wherein step 3. the statistical analysis index is used to evaluate the proposed COVID-19 classification and detection model, and the evaluation index (i.e. accuracy, specificity, sensitivity and recall) is calculated as follows:
the detected image can be finally determined as TP, TN, FP and FN, wherein TP is true positive, TN is true negative, FP is false positive and FN is false negative.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111134598.XA CN113920071A (en) | 2021-09-27 | 2021-09-27 | New coronavirus image identification method based on convolutional neural network algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111134598.XA CN113920071A (en) | 2021-09-27 | 2021-09-27 | New coronavirus image identification method based on convolutional neural network algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113920071A true CN113920071A (en) | 2022-01-11 |
Family
ID=79236211
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111134598.XA Withdrawn CN113920071A (en) | 2021-09-27 | 2021-09-27 | New coronavirus image identification method based on convolutional neural network algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113920071A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115797340A (en) * | 2023-02-03 | 2023-03-14 | 西南石油大学 | Industrial surface defect detection method based on multi-instance learning |
-
2021
- 2021-09-27 CN CN202111134598.XA patent/CN113920071A/en not_active Withdrawn
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115797340A (en) * | 2023-02-03 | 2023-03-14 | 西南石油大学 | Industrial surface defect detection method based on multi-instance learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Hagerty et al. | Deep learning and handcrafted method fusion: higher diagnostic accuracy for melanoma dermoscopy images | |
Jadoon et al. | Three‐class mammogram classification based on descriptive CNN features | |
Hasoon et al. | COVID-19 anomaly detection and classification method based on supervised machine learning of chest X-ray images | |
Javidi et al. | Vessel segmentation and microaneurysm detection using discriminative dictionary learning and sparse representation | |
Ahmmed et al. | Classification of tumors and it stages in brain MRI using support vector machine and artificial neural network | |
Choi et al. | Genetic programming-based feature transform and classification for the automatic detection of pulmonary nodules on computed tomography images | |
Varela-Santos et al. | A new modular neural network approach with fuzzy response integration for lung disease classification based on multiple objective feature optimization in chest X-ray images | |
Alsaade et al. | Developing a recognition system for classifying covid-19 using a convolutional neural network algorithm | |
Attallah | RADIC: A tool for diagnosing COVID-19 from chest CT and X-ray scans using deep learning and quad-radiomics | |
Rahmawaty et al. | Classification of breast ultrasound images based on texture analysis | |
Vo et al. | Multiple kernel learning for classification of diffuse lung disease using HRCT lung images | |
CN113920071A (en) | New coronavirus image identification method based on convolutional neural network algorithm | |
Molachan et al. | Brain Tumor Detection that uses CNN in MRI | |
Genc et al. | Automated abnormality classification of chest radiographs using MobileNetV2 | |
Malathi et al. | Active contour based segmentation and classification for pleura diseases based on Otsu’s thresholding and support vector machine (SVM) | |
Arunmozhi et al. | Automated detection of COVID-19 lesion in lung CT slices with VGG-UNet and handcrafted features | |
CN110837844A (en) | Pancreatic cystic tumor benign and malignant classification method based on CT image dissimilarity characteristics | |
Gudadhe et al. | Classification of Intracranial Hemorrhage CT images for Stroke Analysis with Transformed and Image-based GLCM Features | |
Karsaz | A modified convolutional neural network architecture for diabetic retinopathy screening using SVDD | |
Ain et al. | Mri based glioma detection and classification into low-grade and high-grade gliomas | |
Maghsoudi et al. | Feature based framework to detect diseases, tumor, and bleeding in wireless capsule endoscopy | |
Choudhury et al. | Automated Detection of Central Retinal Vein Occlusion Using Convolutional Neural Network | |
Mahersia et al. | CAD system for lung nodules detection using wavelet-based approach and intelligent classifiers | |
Rathinam et al. | An Adaptive Fuzzy C-Means segmentation and deep learning model for efficient mammogram classification using VGG-Net | |
Devaraj et al. | Early Detection Glaucoma and Stargardt's Disease Using Deep Learning Techniques. |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20220111 |
|
WW01 | Invention patent application withdrawn after publication |