Automated detection of macular hole in optical coherence tomography images using depth-check algorithm

Abstract

Macular hole is a tear or break in the macula. It is located in the center of the retina and affects central vision of aged people. Optical Coherence Tomography (OCT) enables accurate diagnosis of macular hole. Existing algorithms available to detect cysts and retinal layers, but identifying macular hole in an accurate manner is still a missing entity. Hence we propose an automated system for the accurate macular hole detection. The proposed system has six stages in process. The first stage starts with preprocessing the OCT image, then detecting Nerve Fiber Layer (NFL). The detected NFL layer is then processed and depth feature is extracted. Then the macular hole is detected in OCT images using our proposed system. The proposed system is evaluated with the healthy macula and macular hole OCT images. The proposed system is also compared with other machine learning algorithms. By experimentation results, the proposed algorithm provides 94% accuracy in finding macular hole.