A New Approach to Segment Both Main and Peripheral Retinal Vessels Based on Gray-Voting and Gaussian Mixture Model

• DOI: 10.1371/journal.pone.0127748
• Publication Date: 2015-06-05T18:57:22Z
• Authors: Peishan Dai; Hanyuan Luo; Hanwei Sheng; Yali Zhao; Ling Li; Jing Wu; Yuqian Zhao; Kenji Suzuki
• Keywords: Gabor wavelet; Gabor filter

Vessel segmentation in retinal fundus images is a preliminary step to clinical diagnosis for some systemic diseases and eye diseases. The performances of existing methods segmenting small vessels which are usually more importance than the main not satisfactory use. In this paper, we present method both peripheral vessel segmentation. A local gray-level change enhancement algorithm called gray-voting used enhance vessels, while two-dimensional Gabor wavelet extract vessels. We fuse results with 2D-Gabor filter as pre-processing outcome. Gaussian mixture model then clusters from outcome, fragments obtained using another process, complements cluster extraction already performed. At last step, eliminate that do belong based on shape fragments. evaluated approach two publicly available DRIVE (Staal et al., 2004) STARE (Hoover at., 2000) datasets manually segmented results. For dataset, when second include much first "gold standard," achieved an average sensitivity, accuracy specificity 65.0%, 92.1% 97.0%, respectively. sensitivities were higher those other methods, comparable specificities; these thus demonstrated was sensitive detection