CNN with Multiple Inputs for Automatic Glaucoma Assessment Using Fundus Images

2022 ◽  
Author(s):  
Abdelali Elmoufidi ◽  
Ayoub Skouta ◽  
Said Jai-Andaloussi ◽  
Ouail Ouchetto
Keyword(s):  
Principal Component ◽  
Support Vector ◽  
Fundus Images ◽  
Ocular Complications ◽  
Bag Of Features ◽  
Glaucoma Diagnosis ◽  
Mode Decomposition ◽  
Diabetic Proliferative Retinopathy ◽  
Model Training ◽  
Public Datasets

In the area of ophthalmology, glaucoma affects an increasing number of people. It is a major cause of blindness. Early detection avoids severe ocular complications such as glaucoma, cystoid macular edema, or diabetic proliferative retinopathy. Intelligent artificial intelligence has been confirmed beneficial for glaucoma assessment. In this paper, we describe an approach to automate glaucoma diagnosis using funds images. The setup of the proposed framework is in order: The Bi-dimensional Empirical Mode Decomposition (BEMD) algorithm is applied to decompose the Regions of Interest (ROI) to components (BIMFs+residue). CNN architecture VGG19 is implemented to extract features from decomposed BEMD components. Then, we fuse the features of the same ROI in a bag of features. These last very long; therefore, Principal Component Analysis (PCA) are used to reduce features dimensions. The bags of features obtained are the input parameters of the implemented classifier based on the Support Vector Machine (SVM). To train the built models, we have used two public datasets, which are ACRIMA and REFUGE. For testing our models, we have used a part of ACRIMA and REFUGE plus four other public datasets, which are RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF. The overall precision of 98.31%, 98.61%, 96.43%, 96.67%, 95.24%, and 98.60% is obtained on ACRIMA, REFUGE, RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF datasets, respectively, by using the model trained on REFUGE. Again an accuracy of 98.92%, 99.06%, 98.27%, 97.10%, 96.97%, and 96.36% is obtained in the ACRIMA, REFUGE, RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF datasets, respectively, using the model training on ACRIMA. The experimental results obtained from different datasets demonstrate the efficiency and robustness of the proposed approach. A comparison with some recent previous work in the literature has shown a significant advancement in our proposal.

scholarly journals CNN with Multiple Input for automatic glaucoma assessment using Fundus Images

2021 ◽  
Author(s):  
Abdelali ELMOUFIDI ◽  
Said Jai-andaloussi
Keyword(s):  
Principal Component ◽  
Support Vector ◽  
Ocular Complications ◽  
Bag Of Features ◽  
Glaucoma Diagnosis ◽  
Mode Decomposition ◽  
Multiple Input ◽  
Diabetic Proliferative Retinopathy ◽  
Model Training ◽  
Public Datasets

Abstract In the area of ophthalmology, glaucoma affects an increasing number of people. It is a major cause of blindness. Early detection avoids severe ocular complications such as glaucoma, cystoid macular edema, or diabetic proliferative retinopathy. Intelligent artificial has been confirmed beneficial for glaucoma assessment. In this paper, we describe an approach to automate glaucoma diagnosis using funds images. The setup of the proposed framework is, in order: The Bi-dimensional Empirical Mode Decomposition (BEMD) algorithm is applied to decompose the Regions of Interest (ROI) to components (BIMFs + residue). CNN architecture VGG19 is implemented to extract features from decomposed BEMD components. Then, we fuse the features of the same ROI in a bag of features. These last are very long; therefore, Principal Component Analyses (PCA) are used to reduce features dimensions. Obtained bags of features are the input parameters of the implemented classifier based on the Support Vector Machine (SVM). To train the built models, we have used two public datasets, which are ACRIMA and REFUGE. For testing our models, we have used a part of ACRIMA and REFUGE plus four other public datasets, which are RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF. The overall accuracy of 98.31%, 98.61%, 96.43%, 96.67%, 95.24%, and 98.60% are obtained on ACRIMA, REFUGE, RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF datasets, respectively, by using the model trained on REFUGE. Against an accuracy of 98.92%, 99.06%, 98.27%, 97.10%, 96.97%, and 96.36% are obtained on ACRIMA, REFUGE, RIM-ONE, ORIGA-light, Drishti-GS1, and sjchoi86-HRF datasets, respectively, using the model training on ACRIMA. Obtained experimental results from different datasets demonstrate the efficiency and robustness of the proposed approach. A comparison with some recent previous work in the literature has shown a significant advancement in our proposal.

Efficient Feature Descriptor using Gabor Filter and Principal Component Analysis for Glaucoma Diagnosis

2021 ◽  
pp. 6787-6794
Author(s):  
Anisha Rebinth, Dr. S. Mohan Kumar
Keyword(s):  
Principal Component Analysis ◽  
Gabor Filter ◽  
Principal Component ◽  
Component Analysis ◽  
Disease Diagnosis ◽  
Image Features ◽  
Support Vector ◽  
Svm Classifier ◽  
Feature Descriptor ◽  
Glaucoma Diagnosis

An automated Computer Aided Diagnosis (CAD) system for glaucoma diagnosis using fundus images is developed. The various glaucoma image classification schemes using the supervised and unsupervised learning approaches are reviewed. The research paper involves three stages of glaucoma disease diagnosis. First, the pre-processing stage the texture features of the fundus image is recorded with a two-dimensional Gabor filter at various sizes and orientations. The image features are generated using higher order statistical characteristics, and then Principal Component Analysis (PCA) is used to select and reduce the dimension of the image features. For the performance study, the Gabor filter based features are extracted from the RIM-ONE and HRF database images, and then Support Vector Machine (SVM) classifier is used for classification. Final stage utilizes the SVM classifier with the Radial Basis Function (RBF) kernel learning technique for the efficient classification of glaucoma disease with accuracy 90%.

scholarly journals Hyperspectral image classification using support vector machines

2020 ◽  
Vol 9 (4) ◽  
pp. 684
Author(s):  
Jonnadula Dr.J.Harikiran Harikiran
Keyword(s):  
Support Vector Machines ◽  
Image Classification ◽  
Hyperspectral Image ◽  
Principal Component ◽  
Support Vector ◽  
Hyperspectral Image Classification ◽  
Novel Approach ◽  
Mode Decomposition ◽  
Vector Machines ◽  
Bidimensional Empirical Mode Decomposition

In this paper, a novel approach for hyperspectral image classification technique is presented using principal component analysis (PCA), bidimensional empirical mode decomposition (BEMD) and support vector machines (SVM). In this process, using PCA feature extraction technique on Hyperspectral Dataset, the first principal component is extracted. This component is supplied as input to BEMD algorithm, which divides the component into four parts, the first three parts represents intrensic mode functions (IMF) and last part shows the residue. These BIMFs and residue image is further taken as input to the SVM for classification. The results of experiments on two popular datasets of hyperspectral remote sensing scenes represent that the proposed-model offers a competitive analyticalperformance in comparison to some established methods.

scholarly journals A Novel Shearer Cutting State Recognition Method Based on Improved Variational Mode Decomposition and LSSVM with Acoustic Signals

2020 ◽  
Vol 2020 ◽  
pp. 1-16
Author(s):  
Zhongbin Wang ◽  
Bin Liang ◽  
Lei Si ◽  
Kuangwei Tong ◽  
Chao Tan
Keyword(s):  
Search Algorithm ◽  
Optimal Parameter ◽  
Gravitational Search Algorithm ◽  
Principal Component ◽  
Least Square ◽  
Support Vector ◽  
Variational Mode Decomposition ◽  
Intrinsic Mode Functions ◽  
Recognition Method ◽  
Mode Decomposition

The recognition of shearer cutting state is the key technology to realize the intelligent control of the shearer, which has become a highly difficult subject concerned by the world. This paper takes the sound signal as analytic objects and proposes a novel recognition method based on the combination of variational mode decomposition (VMD), principal component analysis method (PCA), and least square support vector machine (LSSVM). VMD can decompose a signal into various modes by using calculus of variation and effectively avoid the false component and mode mixing problems. On this basis, an improved gravitational search algorithm (IGSA) is designed by using the position update mechanism of Levy flight strategy to find the optimal parameter combination of VMD. Then, the feature extraction is achieved by calculating the envelope entropy and kurtosis of the decomposed intrinsic mode functions (IMFs). To avoid dimensional disasters and reinforce the classification performance, PCA is introduced to choose useful features, and the LSSVM-based classifier is reasonably constructed. Finally, the experimental results indicate that the proposed method is more feasible and superior in the recognition of shearer cutting states.

Fine-tuned variational mode decomposition for fault diagnosis of rotary machinery

2019 ◽  
Vol 19 (5) ◽  
pp. 1453-1470
Author(s):  
Ali Dibaj ◽  
Mir Mohammad Ettefagh ◽  
Reza Hassannejad ◽  
Mir Biuok Ehghaghi
Keyword(s):  
Principal Component ◽  
Signal Decomposition ◽  
Support Vector ◽  
Variational Mode Decomposition ◽  
Signal Processing Technique ◽  
Mode Decomposition ◽  
Reconstructed Signal ◽  
Optimal Values ◽  
The Mean ◽  
Mixing Problem

Variational mode decomposition is a powerful signal processing technique that can adaptively decompose a multi-component signal into a number of modes, via solving an optimization problem. The optimal performance of this method in signal decomposition and avoiding of the mode mixing problem strictly relies on the true selection of decomposition parameters, that is, the number of extracted modes ( K) and the mode frequency bandwidth control parameter ( α). In the literature, the optimal values of these parameters are achieved by evaluating fault-related indices like kurtosis, but such an index is inefficient in judging the decomposition of healthy (without fault-related components), low-defective, and high-noise signals. In this research, a novel method called fine-tuned variational mode decomposition is proposed to determine the optimal values of decomposition parameters K and α, by judging the adaptive indices. In this proposed method, the optimal values of these parameters are obtained by minimizing the mean bandwidth of the extracted modes. In order to achieve these optimal values, the mean correlation coefficients between the adjacent modes and the energy loss coefficient between the original signal and the reconstructed signal, should not exceed of defined thresholds for optimal values. The proposed method is applied to the simulation signal and experimental ones collected from the automobile gearbox system. Comparing this method with those in the literature exhibits its higher effectiveness in the true decomposition of signals with different natures. It is also shown that using the proposed method for signal decomposition is able to correctly classify the healthy and defective states of the gearbox system alongside the principal component analysis method and support vector machine classifier.

Bearing degradation state recognition based on kernel PCA and wavelet kernel SVM

2014 ◽  
Vol 229 (15) ◽  
pp. 2827-2834 ◽  
Author(s):  
Shaojiang Dong ◽  
Dihua Sun ◽  
Baoping Tang ◽  
Zhengyuan Gao ◽  
Yingrui Wang ◽  
...  
Keyword(s):  
Principal Component ◽  
Kernel Principal Component Analysis ◽  
Support Vector ◽  
Feature Extraction Method ◽  
State Identification ◽  
Mode Decomposition ◽  
Kernel Support Vector Machine ◽  
Characteristic Features ◽  
Nonlinear Feature Extraction ◽  
Nonlinear Feature

In order to effectively recognize the bearing’s running state, a new method based on kernel principal component analysis (KPCA) and the Morlet wavelet kernel support vector machine (MWSVM) was proposed. First, the gathered vibration signals were decomposed by the empirical mode decomposition (EMD) to obtain the corresponding intrinsic mode function (IMF). The EMD energy entropy that includes dominant fault information is defined as the characteristic features. However, the extracted features remained high-dimensional, and excessive redundant information still existed. Therefore, the nonlinear feature extraction method KPCA was introduced to extract the characteristic features and to reduce the dimension. The extracted characteristic features were inputted into the MWSVM to train and construct the running state identification model, and the bearing’s running state identification was thereby realized. Cases of test and actual were analyzed. The results validate the effectiveness of the proposed algorithm.

scholarly journals Glaucoma diagnosis using discrete wavelet transform and histogram features from fundus images

2018 ◽  
Vol 7 (4) ◽  
pp. 2546 ◽  
Author(s):  
Bhupendra Kirar ◽  
Dheeraj Agrawal
Keyword(s):  
Wavelet Transform ◽  
Discrete Wavelet Transform ◽  
Classification Accuracy ◽  
Cross Validation ◽  
Fluid Pressure ◽  
Nerve Fibers ◽  
Support Vector ◽  
Discrete Wavelet ◽  
Fundus Images ◽  
Glaucoma Diagnosis

Glaucoma is one of the main eye diseases; it cause progressive deterioration of optic nerve fibers due to increased fluid pressure. The existing methods of glaucoma diagnosis are time consuming, expensive and require practiced clinicians to understand the eye problems. Hence fast, cheap and more accurate glaucoma diagnosis methods are needed. This paper presents an innovative idea for diagnosis of glaucoma using third level two dimensional discrete wavelet transform (2D DWT) and histogram features from fundus images. The 2D DWT is used to decompose the glaucoma and healthy images and histogram features are extracted from 2D DWT decomposed sub band images. The least square support vector machine (LS-SVM) is used as a classifier which classifies the glaucoma and healthy images using the extracted features. The proposed method yielded classification accuracy of 88.33%, 87.50%, and 86.67% for ten, eight and fivefold cross validation respectively. The obtained classification accuracy, sensitivity and specificity are 88.33%, 90.00%, and 85.00% for tenfold cross validation respectively. Obtained results prove that the performance of the proposed method is better compared to the existing methods. It may considerably increases the diagnosis speed of ophthalmologists.

scholarly journals AUTOMATED GLAUCOMA DETECTION USING HYBRID FEATURE EXTRACTION IN RETINAL FUNDUS IMAGES

2013 ◽  
Vol 13 (01) ◽  
pp. 1350011 ◽  
Author(s):  
M MUTHU RAMA KRISHNAN ◽  
OLIVER FAUST
Keyword(s):  
Feature Extraction ◽  
Mass Screening ◽  
Risk Index ◽  
Support Vector ◽  
Svm Classifier ◽  
Discrete Wavelet ◽  
Fundus Images ◽  
Glaucoma Diagnosis ◽  
Glaucoma Detection ◽  
Hybrid Feature Extraction

Glaucoma is one of the most common causes of blindness. Robust mass screening may help to extend the symptom-free life for affected patients. To realize mass screening requires a cost-effective glaucoma detection method which integrates well with digital medical and administrative processes. To address these requirements, we propose a novel low cost automated glaucoma diagnosis system based on hybrid feature extraction from digital fundus images. The paper discusses a system for the automated identification of normal and glaucoma classes using higher order spectra (HOS), trace transform (TT), and discrete wavelet transform (DWT) features. The extracted features are fed to a support vector machine (SVM) classifier with linear, polynomial order 1, 2, 3 and radial basis function (RBF) in order to select the best kernel for automated decision making. In this work, the SVM classifier, with a polynomial order 2 kernel function, was able to identify glaucoma and normal images with an accuracy of 91.67%, and sensitivity and specificity of 90% and 93.33%, respectively. Furthermore, we propose a novel integrated index called Glaucoma Risk Index (GRI) which is composed from HOS, TT, and DWT features, to diagnose the unknown class using a single feature. We hope that this GRI will aid clinicians to make a faster glaucoma diagnosis during the mass screening of normal/glaucoma images.

Sign in / Sign up

Export Citation Format

Share Document