Vessel segmentation in 2-D optical coherence tomography images

Abstract

This paper described a novel region segmentation method to avoid difficulties of the threshold process used in traditional segmentation methods in 2-D optical coherence tomography (OCT) images. The speckle effect and diffusion problems make traditional image processing methods such as Canny edge and Otsu methods fail on finding layers and region edges in OCT images. The overcomplete-wavelet-frame-based fractal signature method based on high-pass information and a fuzzy-c-mean algorithm is considered to avoid the threshold processing, but the high-pass information is distorted because of noises and diffusions. To improve the high-pass information distortion problem, the proposed method uses the mean value and an enhanced-fuzzy-c-mean algorithm to cluster pixels in 2-D OCT images and find the edge between different clustered regions. The vessel OCT images are tested in the experiment, and the experimental results show that the proposed method performs with more accurate segmentation results than the overcomplete-wavelet-frame-based fractal signature method.