AI-enabled radar a new tool for diagnosing, monitoring glaucoma
Researchers used thousands of visual fields to develop an artificial intelligence-enabled visualization tool to project the outcomes and the functional worsening of glaucoma, according to a study presented at the American Academy of Ophthalmology annual meeting.
The glaucoma radar can be used for personalized monitoring of visual functional loss, giving patients a tailored look at the trajectory of their vision over time, Siamak Yousefi, PhD, lead developer of the tool and co-author of the study, told Ocular Surgery News.
“Like radar that has a complex system but shows the outcome on a simple screen to track airplanes, the glaucoma radar system is complex but ultimately projects the outcome on a simple and user-friendly screen for tracking visual functional worsening of eyes with glaucoma. It also provides several layers of glaucoma knowledge including information on global severity of visual function, extent of visual functional loss in hemifields and local patterns of visual functional loss,” he said.
Yousefi and colleagues used visual fields from benchmark data sets to validate the tool. The researches acquired 13,231 visual fields, taken from 8,077 subjects, from the Humphrey Field Analyzer (Zeiss). The researchers extracted linear and nonlinear structures using principal component analysis and t-distributed stochastic neighbor (t-SNE) embedding to identify clusters on the t-SNE map.
“We used tens of thousands of visual fields from eyes with glaucoma and normal eyes to identify functional loss characteristics using advanced data mining techniques and then visualized the information on a screen that we called glaucoma radar. Global and local characteristics of visual function loss from past glaucoma patients are retained on the radar and can be used to interpret the functional status of new patients,” Yousefi said.
Using sample eyes from two independent benchmark data sets, which contained confirmed progression or non-progression patients, the researchers consistently verified the visual functional worsening or no worsening on the t-SNE map, Yousefi said.
Possibility of personalized monitoring
The radar potentially provides clinicians with three layers of advanced visualization of glaucoma knowledge, including global visual functional severity, extent of visual functional loss in hemifields and local patterns of visual functional loss, Yousefi said in his AAO presentation.
“Glaucoma radar can be used for personalized monitoring of the visual functional loss of eyes with glaucoma. The trajectory of glaucoma worsening of different patients could be kept on the screen until the next visit of the patient, thus allowing the clinician to see the trajectory of visual worsening at all times. Trajectories of the visual functional worsening of different patients can be simply compared on the glaucoma radar, thus allowing the clinician to identify which patients are worsening fast,” he said.
The potential utility of the glaucoma radar tool is huge, as it can readily provide clinicians with rapid visualization of patients who have little damage or severe loss due to their glaucoma, Louis R. Pasquale, MD, FARVO, co-author of the study and mentor to Yousefi, told Ocular Surgery News.
The radar can be used to better manage and track patients, their disease progression and their current treatment protocol, he said.
“You can look at your radar and say, ‘Look at these patients here. They’re stable, but these patients over here are progressing. Why are they progressing? Do we need to optimize their therapy? Do they have a special genetic or medical risk of their disease? Maybe we should allocate more resources to these patients and less to the patients that are stable so we can optimize care.’ I think that’s a major usefulness for the radar,” he said.
What makes this tool different from other models is that the patient’s outcome is interpretable and can be visualized on a simple screen, Yousefi said.
The radar’s ability to determine vision worsening and progression can let the physician see how the disease is affecting a patient’s quality of life. These factors are beyond most other models that generate only a “simple binary response of whether the patient is progressing or not,” he said.
“While most of the current AI models in the form of deep learning typically generate a black box and provide a yes-no outcome, models like glaucoma radar that provide detailed information may be more helpful in glaucoma clinics and research settings for monitoring patients with glaucoma. We believe glaucoma radar is an example of the next-generation tools for glaucoma assessment that provide multiple layers of glaucoma knowledge on advanced visualization screens,” Yousefi said.
- Yousefi S, et al. Monitoring visual functional worsening in patients with glaucoma using an AI-enabled radar map. Presented at: American Academy of Ophthalmology annual meeting; October 11-15, 2019; San Francisco.
- For more information:
- Louis R. Pasquale, MD, FARVO, can be reached at Mount Sinai Hospital, 1468 Madison Ave., New York, NY 10029; email: firstname.lastname@example.org.
- Siamak Yousefi, PhD, can be reached at University of Tennessee Health Science Center, 930 Madison Ave., Suite #726, Memphis, TN 38163; email: email@example.com.
Disclosures: Pasquale reports no relevant financial disclosures. Yousefi reports receiving funding from the National Eye Institute.