Deep Learning Approaches to Detect Glaucoma and Predict Progression from Spectral Domain Optical Coherence Tomography

Primary open angle glaucoma (POAG) is a leading cause of blindness in the United States and worldwide. It is estimated that over 2.2 million Americans suffer from POAG and that over 130,000 are legally blind from the disease. As the population ages, the number of people with POAG in the United States increased to over 3.3 million in 2020 and is expected to be >110 million worldwide by 2040. POAG is a progressive disease associated with characteristic functional and structural changes that clinicians use to diagnose and monitor the disease. Optical coherence tomography (OCT) and visual field (VF) testing are the clinical standard for measuring the structural (OCT) and functional (VF) changes associated with the development and progression of POAG. Combining data from these sources as well as information about patient demographics, medical history, and clinical measurements is critical to is critical in detecting glaucoma early, identifying signs of progression, and selecting appropriate treatment. Recent progress in AI and deep learning (DL) have provided tools to build predictive multimodal, models that incorporate multiple different types to make predictions. A central hypothesis of this updated research plan is that applying multimodal, longitudinal DL to clinical measurements, VF testing, and OCT imaging will improve the accuracy of predicting progressive structural and functional changes in glaucoma. This updated plan builds on the original research proposal by incorporating new methods and datasets. With this update, there is even greater potential to improve care and preserve vision by helping clinicians tailor glaucoma management to individual patients. This proposal also summarizes the research, training, and career development achievements made the K99 phase of the award. Working with my mentor, Dr. Linda Zangwill, I was able to conduct and publish impactful research during my K99 phase. I was also able to complete the training and career development objectives laid out in the original proposal. During the mentored phase, I helped developed infrastructure and secure access to real-world clinical datasets that will allow me to make immediate progress on the proposed research. This proposed R00 transition will put me in an ideal position to research, publish, mentor, secure funding, and advance my career as an independent investigator.