Nakul Shekhawat, M.D., M.P.H.; Kunal Parikh, PhD; Rama Chellappa, PhD | Johns Hopkins University
Competition Sponsor: US National Academy of Medicine
Awardee Year: 2023
Cataract is a common cause of visual impairment among elderly nursing home residents, who disproportionately suffer from visual impairment due to lack of access to eye care specialists such as ophthalmologists. Cataract surgery leads to dramatic improvements in vision, cognition, and quality of life but lack of access to cataract screening remains a key barrier to cataract surgery among nursing home. We propose to develop a decentralized, de-skilled, semi-automated, smartphone-based platform for screening and diagnosis of cataracts in the nursing home setting. In Aim 1, we will develop hardware to enable in-focus imaging of complex, multiplanar anterior eye segment anatomy and illumination of critical aspects of crystalline lens/cataract anatomy. In Aim 2, will develop software to enable reliable image capture of cataract anatomy by combining images taken from multiple anatomic planes of the eye, adjusting camera exposure across different patient subgroups, automatically detecting pupil centration prior to image capture, and providing healthcare workers with instant feedback on image quality. In Aim 3, will use 1,200 images and linked clinical data collected via our smartphone platform to develop a machine learning algorithm designed to enable healthcare workers to perform real-time, point-of-care diagnosis of cataract in nursing homes. Our algorithm will incorporate several innovations including bias mitigation, domain adaptation, and incorporation of clinical metadata to improve diagnostic performance. The proposed approach to cataract diagnosis has potential to bypass several longstanding constraints and significantly expand access to cataract screening and vision care in the vulnerable elderly nursing home population.