Segmentation and analysis of retinal layers (ILM & RPE) in Optical Coherence Tomography images with Edema

Optical Coherence Tomography (OCT) is a noninvasive technique and depth-resolved imaging modality which is a prominent ophthalmic diagnostic tool. In this paper, an automated segmentation algorithm to detect few intra-retinal layers which are important for Edema detection present in Spectral Domain Optical Coherence Tomography (SDOCT) images is presented. An algorithm for accurate segmentation of intra-retinal layers for normal subjects and patients with edema is discussed. The layers segmented are Inner Limiting Layer (ILM) and Retinal Pigment Epithelium (RPE) layer. The thickness is measured and then based on the thickness the image is classified as edema or non-edema. The accuracy of the algorithm is found to be more than that of the standard edge based segmentation techniques which are more prone to detection of false and disjoint edges. The graph based segmentation is solely based on pixel intensity variation and distance between neighbour pixels. Using the weighing scheme and shortest path search, it eases the task by identifying the neighbourhood pixel having same or similar intensity value and connects it by the path having the least weight. This segmentation method is less prone to noise and the preprocessing step can be considered as optional.