Segmentation of Choroidal Neovascularization in Fundus Fluorescein Angiograms

Abstract

Choroidal neovascularization (CNV) is a common manifestation of age-related macular degeneration (AMD). It is characterized by the growth of abnormal blood vessels in the choroidal layer causing blurring and deterioration of the vision. In late stages, these abnormal vessels can rupture the retinal layers causing complete loss of vision at the affected regions. Determining the CNV size and type in fluorescein angiograms is required for proper treatment and prognosis of the disease. Computer-aided methods for CNV segmentation is needed not only to reduce the burden of manual segmentation but also to reduce inter- and intraobserver variability. In this paper, we present a framework for segmenting CNV lesions based on parametric modeling of the intensity variation in fundus fluorescein angiograms. First, a novel model is proposed to describe the temporal intensity variation at each pixel in image sequences acquired by fluorescein angiography. The set of model parameters at each pixel are used to segment the image into regions of homogeneous parameters. Preliminary results on datasets from 21 patients with Wet-AMD show the potential of the method to segment CNV lesions in close agreement with the manual segmentation.