Vinayaki, V . D.
Deep Retinal Image Analysis and Classification Using Deer Hunting Optimization-Based Tandem Pulse Coupled Neural Network - Vol, 103(6), Dec - New York Springer 2022 - 1909–1916p
Retinal eye diseases lead to vision loss and visual deficiency. Various kinds of human eye diseases are arteriosclerosis, diabetic retinopathy, hypertension, and glaucoma. Diabetic retinopathy is the form of the injured retina which is occurred by diabetes. The improper treatment without proper observations leads to permanent blindness. Therefore, it is necessary to detect the disease at an earlier stage to protect their vision. The main aim of this research is to detect diabetic retinopathy early from the fundus images. The proposed approach detects diabetic retinopathy through four major stages, namely pre-processing, segmentation, feature extraction, and classification. The proposed classifier uses the deep neural network for classifying the diabetic retinopathy (DR) infected images and normal images. Here, the proposed approach utilizes features such as coherence, edge features, shape features, local binary pattern, and Gray Level Co-occurrence Matrix (GLCM) features from the segmented output. The performance measures like sensitivity, specificity, accuracy, precision, F1-measure and processing time are utilized for the estimation of the proposed classifier and compared to other approaches. From the performance evaluation, it is noted that the proposed diabetic retinopathy detection has obtained an accuracy of 99.42%, a sensitivity of 98.84%, and a specificity of 96.49% that outperforming other state-of-art methods.
Electrical Engineering
Deep Retinal Image Analysis and Classification Using Deer Hunting Optimization-Based Tandem Pulse Coupled Neural Network - Vol, 103(6), Dec - New York Springer 2022 - 1909–1916p
Retinal eye diseases lead to vision loss and visual deficiency. Various kinds of human eye diseases are arteriosclerosis, diabetic retinopathy, hypertension, and glaucoma. Diabetic retinopathy is the form of the injured retina which is occurred by diabetes. The improper treatment without proper observations leads to permanent blindness. Therefore, it is necessary to detect the disease at an earlier stage to protect their vision. The main aim of this research is to detect diabetic retinopathy early from the fundus images. The proposed approach detects diabetic retinopathy through four major stages, namely pre-processing, segmentation, feature extraction, and classification. The proposed classifier uses the deep neural network for classifying the diabetic retinopathy (DR) infected images and normal images. Here, the proposed approach utilizes features such as coherence, edge features, shape features, local binary pattern, and Gray Level Co-occurrence Matrix (GLCM) features from the segmented output. The performance measures like sensitivity, specificity, accuracy, precision, F1-measure and processing time are utilized for the estimation of the proposed classifier and compared to other approaches. From the performance evaluation, it is noted that the proposed diabetic retinopathy detection has obtained an accuracy of 99.42%, a sensitivity of 98.84%, and a specificity of 96.49% that outperforming other state-of-art methods.
Electrical Engineering