Interactive Visualization and Manipulation of Volumetric In-Vivo Retinal Images Acquired with Optical Coherence Tomography

This dissertation presents methods for the efficient visualization and manipulation of volumetric data sets. Primarily considered are the challenges presented by in-vivo retinal images acquired through optical coherence tomography (OCT). These challenges include creating a user experience with a low learning curve for clinicians, adapting methods to work within the limited time available in a clinical setting and creating the quality of tools and visualizations needed to make accurate diagnoses and support ophthalmology research. An advanced volume visualization framework that is able to adapt to and fully utilize a broad range of hardware configurations in order to meet any hardware limitations of clinical or research computers while maintaining interactive frame rates is presented. To increase the conveyance of relevant information about a given data set and improve diagnosis, several important visual cues that are specifically targeted to the needs of in vivo retinal images are used to augment standard volume visualization techniques, including advanced lighting and dynamic shadows. Several volumetric annotation techniques that allow the user to add context to the visualization and help bridge the gap between old and new acquisition and visualization techniques are also presented. As OCT acquisition occurs in a curvilinear coordinate system, while most volume visualization techniques are restricted to rectilinear coordinates, additional rendering techniques that provide physically accurate visualizations that preserve data integrity have been developed. A support vector machine (SVM) segmentation approach to quickly and intuitively isolate areas of interest and extract useful metrics is also presented.