Performance Analysis of Convolutional Neural Network for Retinal Image Classification
1Dr.C.R. Dhivyaa, R. Sudhakar, K. Nithya and E. Prabhakar
The performance of convolutional neural network(CNN) has been evaluated and discussed in this paper by comparing the other existing classification techniques. Most of the state-of-the-art classification techniques are trying to detect the abnormal retinal images from the color retinal images. The existing classification techniques misclassify the abnormal retinal image as normal retinal image and it will be required for the diagnosis purpose. In order to overcome the limitations of misclassification, the enhanced convolutional neural network(CNN) is proposed and analyzed to detect the affected retinal images from the color retinal images. The performance of existing classification techniques and proposed classification technique is evaluated and compared for detecting the abnormal images in color retinal images. The proposed CNN provides best result by comparing the experimental results of all the algorithms and it is suitable for detecting the abnormal images in the retinal images.
Retinal Images, Classification, Fuzzy Set, Diabetic Retinopathy (DR), Convolutional Neural Network (CNN)