A Comprehensive Approach for Predicting Different Types of Retinal Detachment with ML Algorithms

The usage of neural networks to image processing has grown in popularity as computer technology and hardware have advanced over time. Soon, DL attracted the attention of healthcare professionals as well, and it began to be utilized to categorize disorders. Several studies are now being conducted to use deep-learning (DL) algorithms to forecast retinal disorders. Meanwhile, very little study has been done on the prediction of DRUSEN, Diabetic Macular Edoema, and Choroid Neo Vascularization (CNV). In this study, we used two DL algorithms Convolutional Neural Network (CNN) and Artificial Neural Network (ANN) to classify Optical Coherence Tomography (OCT) pictures into four categories (CNV, DME, DRUSEN, and natural retina). We’ve used several preprocessing techniques on the photos before sending them to the neural network. Additionally, we have created various models for every method. There are various numbers of hidden layers associated with every model. After conducting our research, proposed that CNN with four hidden layers operate significantly better than any other neural network models and produce 0.86 precision.