The dual-bootstrap iterative closest point algorithm with application to retinal image registration

Abstract

Motivated by the problem of retinal image registration, this paper introduces and analyzes a new registration algorithm called Dual-Bootstrap Iterative Closest Point (Dual-Bootstrap ICP). The approach is to start from one or more initial, low-order estimates that are only accurate in small image regions, called bootstrap regions. In each bootstrap region, the algorithm iteratively: 1) refines the transformation estimate using constraints only from within the bootstrap region; 2) expands the bootstrap region; and 3) tests to see if a higher order transformation model can be used, stopping when the region expands to cover the overlap between images. Steps 1): and 3), the bootstrap steps, are governed by the covariance matrix of the estimated transformation. Estimation refinement [Step 2)] uses a novel robust version of the ICP algorithm. In registering retinal image pairs, Dual-Bootstrap ICP is initialized by automatically matching individual vascular landmarks, and it aligns images based on detected blood vessel centerlines. The resulting quadratic transformations are accurate to less than a pixel. On tests involving approximately 6000 image pairs, it successfully registered 99.5% of the pairs containing at least one common landmark, and 100% of the pairs containing at least one common landmark and at least 35% image overlap.

Comparison of image registration methods for composing spectral retinal images

“…Based on the experiments in [9], a quadratic transformation was deemed appropriate to represent the deformations due to eye movement during retinal image acquisition. The parameter distribution for the transformations was determined from a previously registered true retinal channel image set (with 1 442 images and successful registration con rmed by an expert) by using kernel density estimation.…”

Hybrid Retinal Image Registration Using Mutual Information and Salient Features

“…Also, non-vascular regions of unhealthy retinas exhibit a variety of pathologies over time, which makes the registration difficult. In contrast, vascular regions are relatively stable over time [3]. In recent years a number of retinal image registration methods have been proposed.…”

“…Typically, these methods can be classified into intensity-based, featurebased, or hybrid-based methods. Feature-based methods extract features from a retinal image first, such as vascular bifurcation points [3], whole vasculature [4], and optic disk [5]. Then, the registration process that finds the best transform parameters is performed by maximizing a similarity measure based on correspondences of the extracted features.…”

Accurate Image Registration for Quadrature Tomographic Microscopy

“…Estimation of Θ follows the method in [9], with some modifications, for full automation. The process consists of 2 steps:…”