Automated detection of full thickness macular hole in optical coherence tomography images

Abstract

Macular hole is a break or opening forms in the macula, which commonly affects the central vision of aged people and
vision will be blurry or wavy. Optical Coherence Tomography (OCT) imaging enables accurate diagnosis of full thickness
macular hole. Many algorithms available for retinal layer segmentation, finding intraretinal fluid and automatic cyst
detection, but identification of full thickness macular hole in an automatic manner is still a missing entity. Hence we
proposed an automated system for identifying full thickness macular hole. The proposed system has six modules in process
and starts with pre-processing the OCT image, boundary detection, retinal layer selection, Nerve Fiber Layer (NFL)
segmentation, NFL layer correction and followed by full thickness macular hole detection. The proposed system is
evaluated in OCT images with 50 patients with full thickness macular hole and 50 patients with healthy macula. By
experimentation, it is inferred that the proposed system provides 87% accuracy, 82% sensitivity and 92% specificity.