[Advances in Intelligent Systems and Computing] Intelligent Computing Volume 857 (Proceedings of the 2018 Computing Conference, Volume 2) || Automated Inner Limiting Membrane Segmentation in OCT Retinal Images for Glaucoma Detection

摘要： This paper proposes an automated algorithm to segment inner limiting membrane, i.e., the top most retinal layer in spectral domain optical coherence tomography (SD-OCT) scan. Its segmentation enables ophthalmologist to diagnose retinal diseases like macular edema and glaucoma as they affect ILM layer. The foremost purpose behind this segmentation is to correctly detect and diagnose glaucoma. Glaucoma is declared as the second common cause of blindness by World Health Organization, and can cause severe damage if not treated earlier. Optic-nerve-head (ONH) centered OCT scans are considered for glaucoma. Glaucoma can be described as cupping of optic nerve head, i.e. increase in the diameter of optic cup, and ends up in increasing cup to disc diameters (CDR) ratio. ILM is steeper in central-cup section in glaucoma images than normal images. Therefore, ILM is used to extract cup from macula or ONH centered OCT image volumes and then classify them further as glaucoma-tic or normal eye.