|M.Sc Student||Meitav Nizan|
|Subject||High Resolution Retinal Imaging and Pattern Analysis without|
|Department||Department of Physics||Supervisor||Dr. Erez Ribak|
|Full Thesis text|
High-resolution retinal imaging requires dilating the pupil, and therefore exposing aberrations that blur the image. To regain the high resolution, adaptive optics is employed to minimize these aberrations. The required technology is sensitive to the eye's spatial and temporal changes and as a result is complex and expensive.
In this work, a non-adaptive optics high resolution imaging technique was developed by taking advantage of the natural movement of the eye, to average out some of the aberrations. By constructing a high-resolution fundus camera, and developing a corresponding image processing method, structures of the size of single cells in the central and peripheral part of the living human retina were resolvable.
Being the closest packing possible, the hexagonal packing of photoreceptors sets the upper limit of the sampling frequency, namely constituting a criterion for visual acuity. In this work two independent automated methods for measuring the average size and fraction of hexagonal packed photoreceptors were developed. Applying these two techniques on images, obtained by the aforementioned method, it is possible the measure hexagonal packed cones at the central and peripheral retinal regions.
The results of this analysis had proven that the hexagonal density is mostly set by adjacent cones at different areas, and is decreasing with eccentricity. In addition to that, it was found that there are local areas of high density of hexagons with respect to their surroundings. These findings imply that high spatial frequencies are also being sampled in peripheral parts of the retina while the randomly arranged cones are averaging out frequencies above the Nyquist frequency.
An attempt to improve even further the resolution of the retinal image was done by building a new apparatus based on corneal near annulment. This was done by immersing the eye in custom designed goggles filled with water. In addition, the inner ocular aberrations were reduced by using an aspheric lens, which was specially optimized and made for this purpose. The optical optimization was based on a schematic eye model that was developed by Dr. A. Goncharov from Applied Optics group at National University of Ireland, Galway, giving a theoretical resolving limit of approximately two microns. In comparison to the non immersion apparatus final images, there is approximately 20% resolution increase using the goggle-based setup, thus increasing the resolving limit from approximately 4.2 to 3.5 ?m. By further resolving some of the existing difficulties in the goggle-based apparatus a better resolution can be expected.