Research Project
6884316
Retinal images remain an important diagnostic tool in ophthalmology. Today 35mm color images remain the "gold standard" by which new digital cameras are compared and judged. Over the past several years the number of pixels available to the digital camera has continued to increase to a point where the pixel footprint on the retina now closely matches the resolution that is allowed by the human eye. Further improvements in spatial resolution will require systems, such as adaptive optics cameras, or approaches, such as image deconvolution, to compensate for aberrations introduced primarily by the anterior segment of the eye. This application will demonstrate a low-cost approach for improving image quality, contrast and resolution. Further, we will show that the improved image quality will have major medical significance by allowing improved detection of lesions, such as found in diabetic retinopathy. Application of advance imaging technologies such as wavefront sensing of ocular aberrations and correction of these aberrations with adaptive optics (AO) can significantly increase resolution and improve image quality over the current fundus images. However, AO systems are currently too expensive to be viable for clinical implementation. The objective of the project is to demonstrate that significant information can be provided to the clinical through image deconvolution, where the ocular aberrations are measured by a wavefront sensor and applied to the image deconvolution. The pilot study shows that through the application of wavefront-based and blind image deconvolution, image quality has been improved significantly. The nerve fiber layer and important pathologies such as micro-aneurysms, hemorrhages, and neovascularization that are not observable in standard retinal images or seen by ophthalmologist in the direct biomicroscopy examination are made visible, providing critical information for early and accurate diagnosis of the diseases. The objective of Phase I is to develop and optimize deconvolution techniques that are efficient and require no human interaction. Objective quantitative image quality measurements will be selected for determination of improvement in image quality after restoration by deconvolution. Phase I will quantitate the degree of improvement provided by deconvolution with wavefront sensing.
