Improved OCT Human Corneal segmentation Using Bayesian Residual Transform

The inherent poor signal to noise ratio of Optical Coherent Tomography (OCT) is considered as a main limitation of OCT segmentation, particularly because images are sampled quickly, at high resolutions, and in-vivo. Furthermore, speckle noise is generated by the reflections of the OCT LASER limits the ability of automatically segmenting OCT images. This paper presents a novel method to automatically segment human corneal OCT images. The proposed method uses Bayesian Residual Transform (BRT) to build a noise robust external force map, that guides active contours model to the corneal data in OCT images. Experimental results show that the proposed method outperforms the classical as well as the state-ofthe- art methods.