Analysis of shape assumptions in 3D reconstruction of retina from multiple fundus images

Abstract

Utilizing priors about the shape of retinal surface is important for accurate reconstruction. We present a detailed analysis of geometrical shape priors in the 3D reconstruction of retina. We first approximate the retinal surface either as a sphere inspired by the actual shape of the eyeball, or as a plane inspired by the 2D mosaicing approaches. Based on this approximation, we perform an initial camera localization with a 2D-to-3D registration procedure. Then, parameters of the surface and the camera poses are refined through a nonlinear least squares optimization using different shape priors. The resulting 3D model and camera poses can be used for intuitively visualizing the retinal images with a model-guided browsing interface.

Geometrically constrained sub-pixel disparity estimation from stereo images of the retinal fundus

“…A point m(u, v) can be mapped back to M(x, y, z) by computing the line of sight using M −1 int and the intersection with the fundus surface defined in equation (1).…”

“…Retinal image processing is a well-studied field and several studies about depth estimation of retinal fundus exists [1,2,3,4].…”

Design of optical system for binocular fundus camera

“…Fundus camera meets the actual needs of the majority of patients. Local three-dimensional reconstruction of glaucomatous optic disk was proposed by Corona et al [7] and external three-reconstruction of fundus based on actual shape of retina was studied by Ataer-Cansizoglu et al [8] However, they both achieved three-dimensional reconstruction using multiple images registration and their effects are not ideal. The main reason is that living human's eye may produce tremors, drifts, micro saccades and other physiological phenomena, which lead to higher order aberration [9].…”