Enhancing retinal images by nonlinear registration

TL;DR

This paper introduces a nonlinear registration method inspired by solar astrophysics to enhance high-resolution retinal imaging, aiming to improve medical diagnosis by detecting abnormal features and tracking structural changes.

Contribution

It applies solar astronomy correlation techniques to retinal imaging, enabling better image registration, expanded field of view, and movement analysis for medical diagnostics.

Findings

Enhanced retinal image quality and field of view.

Effective tracking of structural and motion changes.

Potential for improved diagnosis of retinal abnormalities.

Abstract

Being able to image the human retina in high resolution opens a new era in many important fields, such as pharmacological research for retinal diseases, researches in human cognition, nervous system, metabolism and blood stream, to name a few. In this paper, we propose to share the knowledge acquired in the fields of optics and imaging in solar astrophysics in order to improve the retinal imaging at very high spatial resolution in the perspective to perform a medical diagnosis. The main purpose would be to assist health care practitioners by enhancing retinal images and detect abnormal features. We apply a nonlinear registration method using local correlation tracking to increase the field of view and follow structure evolutions using correlation techniques borrowed from solar astronomy technique expertise. Another purpose is to define the tracer of movements after analyzing local correlations to follow the proper motions of an image from one moment to another, such as changes in optical flows that would be of high interest in a medical diagnosis.