Pre-processing, Registration and Quality Assessment of Adaptive Optics Assisted Retinal Images
MetadataShow full item record
This item's downloads: 220 (view details)
In recent decades, adaptive optics (AO) technology has been embedded into retinal imaging devices, producing a new generation of instrument, which can provide retinal images with cellular resolution. This new technological advancement allows viewing of retinal microscopic structure, which is of great significance for the diagnosis, and subsequent treatment monitoring, of retinal pathologies that can result in visual loss. Developing compact and simplified AO assisted retinal imaging devices with an automated feature analysis is the current focus of interest for transferring AO technology to clinical use. In this study, we present an enhanced processing of sequences of retinal images obtained using an AO flood illumination system. We aim to provide image processing techniques for pre-processing, assessing the quality and image registration of cone photoreceptor and retinal nerve fiber layer (RNFL) images. Our results demonstrate the effectiveness of a wavelet based approach to correcting uneven illumination and automatic evaluation of image quality in terms of the results of image registration. In particular, we present the significance of image quality analysis while selecting a certain percentage of the sharpest images in a sequence for image registration. In order to register the images, we include methods that are specifically developed to measure tiny rotations in addition with correlation based techniques to correct for translational motion. We show that correcting for small rotations exhibits a significant improvement, especially at the edges of the image, which is important for creating larger mosaics. We then present the methods of investigating the characteristics of retinal nerve fiber bundles to discriminate RNFL images with good and poor striation. This enables feature comparison between healthy and glaucoma eyes. Finally, we discuss the implications of our results and possible future studies.
This item is available under the Attribution-NonCommercial-NoDerivs 3.0 Ireland. No item may be reproduced for commercial purposes. Please refer to the publisher's URL where this is made available, or to notes contained in the item itself. Other terms may apply.
The following license files are associated with this item: