Temporal registration for low-quality retinal images of the murine eye
1Department of Electrical and Electronic Engineering, University of Bristol, Merchant Venturers Building, Woodland Road, BS8 1UB, UK. l.andreou06@bristol.ac.uk
Abstract:
This paper presents an investigation into different approaches for segmentation-driven retinal image registration. This constitutes an intermediate step towards detecting changes occurring in the topography of blood vessels, which are caused by disease progression. A temporal dataset of retinal images was collected from small animals (i.e. mice). The perceived low quality of the dataset employed favoured the implementation of a simple registration approach that can cope with rotation, translation and scaling, in the presence of major vascular dissimilarities, distortions, noise, and blurring effects. The proposed approach uses a single control point, i.e. the centroid of the optic disc, and achieves accurate registration by matching points in the pair of input images using mean squared error calculation. A number of alternative, more sophisticated methods have been explored alongside the proposed one. While these other methods could prove valuable and perform reasonably well when applied on good quality images, they generally fail when using the dataset at hand.
Insights
A simple retinal image registration method using the optic disc centroid effectively aligns low-quality images for disease progression analysis. This approach overcomes common image artifacts, unlike complex methods.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computational Biology
Background:
- Accurate retinal image registration is crucial for monitoring disease progression by detecting changes in blood vessel topography.
- Low-quality retinal image datasets present significant challenges for traditional registration techniques.
Purpose of the Study:
- To investigate and propose a robust segmentation-driven retinal image registration method suitable for low-quality temporal datasets.
- To develop a registration approach that can handle rotation, translation, scaling, vascular dissimilarities, distortions, noise, and blurring.
Main Methods:
- A simple registration approach utilizing the optic disc centroid as a single control point.
- Matching corresponding points between image pairs using mean squared error (MSE) calculation.
- Comparison with alternative, more sophisticated registration methods on a challenging dataset.
Main Results:
- The proposed simple registration method achieved accurate alignment of low-quality retinal images.
- Complex registration methods explored in the study generally failed when applied to the low-quality dataset.
- The optic disc centroid proved to be an effective control point for registration.
Conclusions:
- A straightforward optic disc centroid-based registration method is effective for low-quality retinal images.
- This approach facilitates the detection of disease-related changes in retinal vasculature.
- The method offers a practical solution for longitudinal studies with challenging imaging conditions.
