Automated Binary and Multiclass Classification of Diabetic Retinopathy Using Haralick and Multiresolution Features

S Gayathri,Varun P Gopi,Adithya K Krishna,P Palanisamy

doi:10.1109/access.2020.2979753

Abstract

Diabetic Retinopathy (DR) is considered as the complication of Diabetes Mellitus that damages the blood vessels in the retina. This is characterized as a serious vision-threatening problem in most of the diabetic subjects. Effective automatic classification of diabetic retinopathy is a challenging task in the medical field. The feature extraction plays an eminent role in the effective classification of disease. The proposed work focuses on the extraction of Haralick and Anisotropic Dual-Tree Complex Wavelet Transform (ADTCWT) features that can perform reliable DR classification from retinal fundus images. The Haralick features are based on second-order statistics and ADTCWT reliably extracts the directional features in images. The proposed work concentrates on both binary classification as well as multiclass classification of DR. The system is evaluated across various classifiers such as Support Vector Machine (SVM), Random Forest, Random Tree, J48 classifiers by giving input image features extracted from the MESSIDOR, KAGGLE and DIARETDB0 databases. The performances of the classifiers are analyzed by comparing specificity, precision, recall, False Positive Rate (FPR) and accuracy values for each classifier. The evaluation results show that by applying the proposed feature extraction method, Random Forest outperforms all the other classifiers with an average accuracy of 99.7% and 99.82% for binary and multiclass classification respectively.

Highlights

Diabetic Retinopathy (DR) is considered as the complication of diabetes mellites
Since Grey Level Co-occurrence Matrix (GLCM) is a powerful tool for texture analysis, and Anisotropic Dual-Tree Complex Wavelet Transform (ADTCWT) provides the benefit of multidirectionality and shift invariance, this hybrid feature extraction method outputs all the required features from the input images
In [5] the haralick feature extraction alone with Support Vector Machine (SVM) classifier produced an accuracy of 84% for DIARETDB0 database and 86% for High-Resolution Fundus (HRF) images, while the combination of haralick and ADTCWT feature extraction with SVM classifier in the proposed method offers a promising DR classification system with an average accuracy of 98.625% and the proposed system works best using random forest classifier with an accuracy of 99.84%

Summary

INTRODUCTION

Diabetic Retinopathy (DR) is considered as the complication of diabetes mellites. The cause of the disease is that for a subject with the history of diabetes mellites, their blood vessels in the retina gradually gets damaged. S. Gayathri et al.: Automated Binary and Multiclass Classification of DR Using Haralick and Multiresolution Features methods developed for its earlier detection still the detection of DR and its severity grading standing as a challenging task. Local features of retinal images are extracted using Local Binary Patterns (LBP) in [9] It is evaluated across Artificial Neural Network (ANN), Random Forest and SVM for the detection task. The proposed work aims to implement an automated system that can perform binary classification(DR or normal classification) as well as multiclass classification (mild DR, moderate DR or severe DR) of DR from retinal fundus images for better disease diagnosis. In order to achieve the goal, a new combined feature extraction technique is implemented using Haralick and ADTCWT methods. The combination of two methods anticipates a good feature extractor for DR classification

METHODOLOGY

PERFORMANCE ANALYSIS AND EVALUATION

Findings

CONCLUSION