Segmentation of retinal cysts from Optical Coherence Tomography volumes via selective enhancement

Automated and accurate segmentation of cystoid structures in Optical\nCoherence Tomography (OCT) is of interest in the early detection of retinal\ndiseases. It is, however, a challenging task. We propose a novel method for\nlocalizing cysts in 3D OCT volumes. The proposed work is biologically inspired\nand based on selective enhancement of the cysts, by inducing motion to a given\nOCT slice. A Convolutional Neural Network (CNN) is designed to learn a mapping\nfunction that combines the result of multiple such motions to produce a\nprobability map for cyst locations in a given slice. The final segmentation of\ncysts is obtained via simple clustering of the detected cyst locations. The\nproposed method is evaluated on two public datasets and one private dataset.\nThe public datasets include the one released for the OPTIMA Cyst segmentation\nchallenge (OCSC) in MICCAI 2015 and the DME dataset. After training on the OCSC\ntrain set, the method achieves a mean Dice Coefficient (DC) of 0.71 on the OCSC\ntest set. The robustness of the algorithm was examined by cross-validation on\nthe DME and AEI (private) datasets and a mean DC values obtained were 0.69 and\n0.79, respectively. Overall, the proposed system outperforms all benchmarks.\nThese results underscore the strengths of the proposed method in handling\nvariations in both data acquisition protocols and scanners.\n