Adaptive Optics Retinal Imaging
National Eye Institute
Investigators
Linked publications, trials & patents
Abstract
By the time diseases of the retina are detected, serious damage has often already been done. An advanced optical imaging instrument utilizing adaptive optics can be used to directly visualize the cellular structure of the retina in the living human eye. Adaptive optics is a technology for measuring and correcting the optical imperfections in the human eye. When adaptive optics is combined with an imaging platform, highly detailed images of the human retina can be acquired. Our research utilizes this technology to image cells in patients eyes through the Adaptive Optics Clinic within the NIH Clinical Center. Processing of adaptive optics is highly time-consuming and labor intensive. We are continuing to pioneer artificial intelligence-based methods for the analysis of adaptive optics imaging data, including for cone photoreceptors and retinal pigment epithelial cells. Evaluating the status of these cells in the living human eye is an important step towards understanding the pathophysiology of disease. We have developed several innovative approaches in the past year, including incorporation of spatial awareness into regression techniques for detection of fluorescently-labeled retinal pigment epithelial cells, as well as the use of generative adversarial networks to synthesize artificial realistic data to improve cone segmentation. This second approach also reduces the amount of data annotation that is needed to establish large training datasets for training deep learning models. We continue to make software tools publicly-available through the NEI Commons. Activities to improve imaging technology are also ongoing, through the design and implementation of upgrades to our state-of-the-art, custom-built adaptive optics instrument in the NEI eye clinic with the overarching goal of augmenting the translational research capabilities at the NIH Clinical Center. A notable strategy developed in the past year is to use ring illumination in combination with sub-Airy disk detection schemes to surpass the optical limit of resolution in the living human eye. This technique improves the clarity with which the smallest cone and rod photoreceptors can be visualized, an important advance for studying microscopic changes due to disease. Through collaboration with Drs. Hammer and Liu at the FDA, we have successfully obtained multimodal images of the retinal pigment epithelial cells using four different modalities, including state-of-the-art adaptive optics optical coherence tomography. This integrated instrument is poised for clinical investigations that are planned and ongoing. Through collaboration with NIH clinicians, we continue to explore how the cells in the eye are affected by both rare genetic eye diseases as well as other retinal diseases. One example of a rare condition is oligocone trichromacy. Using multimodal adaptive optics imaging, we demonstrated for the first time that there are extensive dark cones present in a patient with oligocone trichromacy. Despite the presence of these dark cones, there is still retained visual function, a finding that has important implications establishing the clinical meaning of dark cones, which are being identified in an increasing number of disorders in recent years. These and other advances represent progress toward our overall goal to develop more sensitive methods to monitor the status and progression of disease at the cellular level in patients.
View original record on NIH RePORTER →