Ocular Disease Prediction Using Feature Maps with Convolutional Neural Network (CNN) Method

The most essential and sensory organ for humans is eye whereas degenerative conditions as well as diseased eyes a great concern that effect function of the significant organ. These diseases or issues can frequently be immediately cured or significantly alleviated with accurate early diagnosis by professionals. At present, eye disease diagnosis is labor-intensive, prone to error, and challenging. Diagnosing ocular disease has influences Machine Learning (ML) which is very likely for human eyes from disorders such as Diabetic Retinopathy (DR), glaucoma, cataract, myopia etc., beyond the time limit. Image classification is a type of image processing that assist to understand various features through images. This research focuses feature mapping which act as an interactive classification on Deep Learning (DL) that assists in ocular pathological studies involved automatic recognition of diseases from fundus images. Dataset considered research involves 6392 images from 5000 patients for identifying different ocular diseases. Moreover, feature mapping is used in identifying the best feature set for the available eye disease dataset. Hence, Modified Convolution Neural Network (MCNN) has involved with various sequence to involve feature mapping through internal representation of certain inputs in every convolutional layer of the model. Thus, 97.39% accuracy of MCNN model obtained in right eye help in predicting eye disease in earlier stage to avoid complete blindness.