Classification Based Quantitative Assessment of Cup-to-Disk Ratio in Spectral Domain Optical Coherence Tomography Images

In the process of glaucoma detection, spectral domain optical coherence tomography（SD-OCT） images are better than fundus images in aspects of precision and reliability. To overcome the defects of low accuracy in cup-to-disk ratio assessment based on fundus images, this paper presents a novel method to detect retinal pigment epithelium（RPE） breakpoints in SD-OCT images and to effectively calculate the cup-to-disk ratio. Firstly, we train a classifier with principal component analysis and support vector machine, and then constrain an area where the breakpoints of RPE are searched using SD-OCT and project images. After recognition, the breakpoints are corrected with a label matrix. Finally, the cup-to-disk ratio is calculated based on the locations of the breakpoints. Experimental results demonstrate that the proposed method is effective to detect RPE breakpoints and calculate the cup-to-disk ratio.