21 Oct Glaucoma Visual Functional Progression Can Be Monitored Using AI-Enabled Radar
MedicalResearch.com Interview with:
Louis R. Pasquale, MD
Professor of Ophthalmology
Icahn School of Medicine at Mount Sinai;
Site Chair of the Department of
Ophthalmology at The Mount Sinai Hospital and Mount Sinai Queens;
Vice Chair of Translational Ophthalmology Research
Mount Sinai Health System
MedicalResearch.com: What is the background for this study? What are the main findings?
Response: Individual visual field tests provide a 52-point array of functional information about a glaucoma patient but it does not give us a handle on how functionally disabled they might be. A series of visual field tests need to be assessed for functional progression but current conventional algorithms for doing so are governed by ad hoc rules and the various algorithms available for assessing progression do not agree with one another. Finally, in managed care setting where one might be responsible for allocating resources for large numbers of glaucoma patients, it would be valuable to quickly visualize which patients are progressing rapidly and which ones are stable. This could allow for proper allocation of resources and perhaps inquiry into why a subset of patients are doing poorly.
We wanted to develop an easy to use tool to quickly visualize how individual glaucoma patients and how groups of glaucoma patients are doing from a functional perspective.
MedicalResearch.com: What is meant by AI Enabled radar?
Response: The radar is a 2-dimensional map that depicts the full spectrum of functional variation exhibited by glaucoma patients. Rather than manually and arbitrarily creating such a map, we decided to use unsupervised computer algorithms to construct it. Specifically we used over 13,000 visual fields and applied principal component analysis to discover the major sources of linear variance in those tests. Then we applied t-distributed stochastic neighbor embedding, a method to discover the nonlinear variation in the dataset that also allowed us to create a 2-dimensional cloud of visual field data. Finally we applied a clustering method to discover aggregates of visual fields that clustered together on the 2-dimensional array of visual fields.
MedicalResearch.com: What are the main findings?
Response: Interrogating the structure of the AI-enabled radar, revealed that it contained 32 clusters and 3 layers of useful information. First, the clusters were arrayed from lower disease severity to higher severity as one courses from upper right to lower left on the radar. Second different regions of the radar highlighted clusters that preferentially involved either the superior hemifield or inferior hemifield. Third, some clusters contain patients with preferential patterns of loss. For example in the lower left region existed a cluster of patients with predominantly superior paracentral visual field loss. Patients with this pattern of loss may have particular difficultly with reading.
The radar was useful for visualizing progressive and non progressive changes in visual fields over time.
MedicalResearch.com: What should readers take away from your report?
Response: Just as one might familiarize oneself with the features of a visual field plot which informs us about the anatomy of the island of vision, it will become useful for eye care providers to become familiar with the 2-dimensional AI enabled radar. It will quickly inform the practitioner about the functional status of a glaucoma patient or a group of glaucoma patients.
MedicalResearch.com: What recommendations do you have for future research as a result of this work?
Response: We are in the process of further validation of the 2-dimensional radar. Specifically we have a set of visual fields for which there is consensus of progression on 4 algorithms. We suspect when these visual fields are projected on the radar the majority of them will show progression. By comparing them to another data set where multiple visual fields were acquired over a short time period, we hope to derive a more objective definition of visual field progression.
Abstract presented at the 2019 AAO meeting
Monitoring Visual Functional Worsening in Patients With Glaucoma
Using an AI-Enabled Radar
The information on MedicalResearch.com is provided for educational purposes only, and is in no way intended to diagnose, cure, or treat any medical or other condition. Always seek the advice of your physician or other qualified health and ask your doctor any questions you may have regarding a medical condition. In addition to all other limitations and disclaimers in this agreement, service provider and its third party providers disclaim any liability or loss in connection with the content provided on this website.
Last Updated on October 21, 2019 by Marie Benz MD FAAD