Optical coherence tomography with the best deep properties for finding age-related macular degeneration

Background An ocular ailment impacts the entirety of sensory functioning, and an undetected and untreated ocular ailment might result in visual impairment. The objective of this study was to create an automated system for detecting age-related macular degeneration (ARMD) using deep-learning (DL) method that combines optimal deep features selected using the arithmetic optimizer (AO). Methods The main stages of the proposed scheme include; image collection from the chosen dataset and resizing, feature extraction using the DL scheme, feature optimization with AO and serial features fusion, and bi-level data classification and five-fold cross validation. The performance of the developed DL system is verified using conventional-, optimal-, and fused-features and the importance of these methods are established based on the chosen performance metrics. Results This study is executed using 1200 numbers of optical coherence tomography images of Normal/ARMD class and the scheme helps in achieving accuracy of >90% and >91% with conventional- and optimal features. Further, the fused-features based classification provided an accuracy of >99% with the support vector machin. Conclusions The achieved results of this study confirm the importance of the developed procedure on the examination of retinal optical coherence tomography images.