A Novel Approach to Segment OCT layers using Graph method for Glaucoma Diagnosis

Optical Coherence Tomography (OCT) is a relatively new non-invasive, cross-sectional imaging technology that describes the internal structures of the retina. The thickness of various retinal layers such as Internal Limiting Membrane (ILM), Retinal Nerve Fibre Layer (RNFL) and Retinal Pigment Epithelium (RPE) in OCT are imperative in the diagnosis of different ophthalmic disorders like Glaucoma. For medical examiners, segmentation of images is a cumbersome task and demands expertise. This paper proposes an algorithm that can assist in segmenting the layers in an OCT image for glaucoma examination. Graph cut method shows promising results in OCT segmentation when compared to conventional algorithms. The proposed algorithm employs a graph cut based OCT segmentation schema, with Dijkstra algorithm using min-heap to reduce the computation time. The algorithm also effectively addresses blood vessel discontinuity with the help of adjacency matrix.