AI-Driven Tools for Automated Design of Higher Order Aberration Optics in Contact Lenses

PROJECT SUMMARY
The Ovitz Corporation has developed a commercial aberrometer (the Ovitz xwave) along with a proprietary
algorithm that is used to design of scleral lenses with precise HOA optics. Despite these innovations, cost of
specialty lenses and time associated with iterative designs proves to be a barrier to treatment. In this proposal,
we propose to we propose to reduce the design time by up to 40 minutes per lens by automating the lens design
process using a machine learning (ML) model. Additionally, we aim to implement our algorithm in a way that
could eliminate at least one fitting session for patients. These efforts will result in lower costs and time
commitments for patients, additionally increasing the patient volume .
The project is designed to train and validate the ML program and examine future pathways to ease the patient’s
treatment pathway. Specifically, we will: 1) Prepare the training dataset for ML by curating our existing data,
which has 28,000 images collected by the xwave system for training ML models. Key curation will occur to
understand the displacement for wavefront images and associated Zernike coefficients. This goal will also
analyze the frequency of lens fitting revisions and associated root cause for patients, in order to benchmark the
performance of the ML and set patient expectations. 2) The second goal will be to train 70% of the dataset on
the curated data, using the remaining 30% to validate the predictive capabilities. 3) The final aim will assess the
algorithm’s ability to determine the lens position on the eye without the need for an additional lens with fiducial
markings, reducing a fitting iteration for patients.

Grant Number: 1R43EY037613-01
NIH Institute/Center: NIH
Principal Investigator: Nicolas Brown