High-resolution retinal imaging can significantly improve the quality of diagnosis, disease progression tracking and assessment of therapy in a broad range of retinal diseases, including retinal degenerations (retinitis pigmentosa), macular telangiectasis, macular dystrophies, age-related macular degeneration (AMD), and inflammatory diseases. Some of these diseases are prevalent (AMD afflicts 12% of the population aged 80+, and retinitis pigmentosa is the most common leading cause of blindness/low-vision in adults 20-60 yrs old) and progress slowly; thereby driving the need for a cost-effective imaging solution that can be broadly deployed for screening and tracking purposes. The availability of new therapeutics further drives this need as the ability to discern the exact impact of the drugs at th cellular level is highly useful in determining treatments. The transverse resolution of conventional retinal imaging systems (i.e. in the plane of the retina), however, is generally limited to around 20 microns due to intrinsic aberrations of the ocular media (in particular the lens and cornea). To overcome this limitation and achieve cellular level resolutions, adaptive optics techniques have been introduced to ophthalmic systems so that a highly-focused spot of light can be raster-scanned directly onto the retina. Although the speed, resolution and capabilities of these systems have steadily improved over the past 10-15 years, very few AO systems have become commercially available, and the systems are not yet sufficiently user-friendly or cost-effective. In ophthalmological clinics, the adaptive optics approach for imaging has had little penetration. Here we propose to explore an entirely different strategy for accomplishing low-cost ultra-high resolution retinal imaging. We term this approach Fourier Ptychographic Retinal Imaging (FPRI). The method simply uses a specially built camera to acquire a sequence of restricted aperture images of the illuminated retina. The aberrative distortions inherent in these images are invariant and can be addressed and corrected post-image- acquisition to render an ultra-high resolution undistorted retinal image by using a new class of optical phase retrieval method called Fourier Ptychography (FP). This approach is highly novel. In comparison to adaptive optics methods which attempt to physically correct aberrations before image acquisition, FPRI will instead simply collect a sequence of raw images and computationally correct the aberrations after image collection. Put in another way, the FPRI approach recasts a problem that is very challenging to solve in the physical realm into a computational problem that can be easily solved in the computational realm.
This proposed project aims to develop a method for performing high resolution retinal imaging by acquiring a sequence of retinal images and applying a new processing method to find and computationally nullify the optical aberrations inherent in the eye. Through this means, we will be able to generate a high resolution retinal image. We believe that this simple and cost-effective approach will be very useful as a clinical tool, in place of the far more complex adaptive optics based ophthalmic imaging methods that have difficulty penetrating the clinical setting.
