The primary commercial objective of this research is to offer a profitable product for retinal specialists that will provide improved 2D visualization of shallow and deep structures in the fundus, using fluorescent dyes. Such fluorescent dyes primarily include indocyanine green (ICG), which provides viewing of deep structures such as choroidal vasculature, and fluorescein (FITC) which provides viewing of retinal vascular and other retinal structures. The focus of this Phase I research will be on ICG imagery. The technology to be developed will be in the form of a software algorithm, with an end product ultimately installed on a Pentium computer platform for marketability. The design of the algorithm will be founded fundamentally on the model-based approach of maximum likelihood estimation. Our experience has been that model-based approaches remove the bulk of the guess work and over- whelming, manpower consuming trial-and-error in formulating a working algorithm design. On the other hand, our experience has been that pure model-based approaches in general are difficult to design profitable products around, because of speed and cost issues, so the algorithm will additionally have non-model-based components in its underlying formulation as well, in order to ultimately provide a working, fast, inexpensive, robust, appealing product.
The specific aims of this Phase I research are to: (1) Collect a number of image data sets from subjects as specified later under Data Collection. (2) Mathematically derive, develop and implement a prototype algorithm. (3) Execute this algorithm on the collected data sets. (4) Evaluate the resulting deconvolved images for resolution improvement and clarity as later specified under Presentation and Evaluation .
The product will be an add-on to Scanning Laser Ophthalmoscopes (SLO s) from several manufacturers. At least 700 SLO sites exist already, and this number is growing rapidly. Worldwide there are at least 5,000 potential sites, assuming 50 percent of retinal specialists will purchase this software.
