The role of quantitative ultrawidefield angiography and machine learning

(Image Credit: AdobeStock/RFBSIP)

With new and emerging technologies in quantitative ultrawidefield angiography (UWFA), Justis P. Ehlers, MD, believes that patients with diabetic eye disease can be treated more effectively by varying treatment approaches.

“To think that the treatment approach should be identical between 2 eyes may be incorrect. Helping to better understand how to identify those patients who will benefit from the best treatment methodology is one of the goals of our research,” Ehlers said. He is the Norman C. and Donna L. Harbert Endowed Chair for Ophthalmic Research at Cleveland Clinic Cole Eye Institute.

During a presentation at the Fourth Annual Cleveland Eye Bank Foundation Virtual Vision Research Symposium,¹ Ehlers offered insights into the value of quantitative UWFA in diabetic retinopathy. He reviewed machine learning-enhanced techniques to quantify diabetic retinopathy features, evaluating their link to disease severity and risk of progression.

Ehlers and his team at Cleveland Clinic Cole Eye Institute have integrated quantitative UWFA with artificial intelligence and machine learning. By capturing images of the entire retina, they assess diabetic retinopathy and disease burden, as well as conditions such as ischemic leakage and microaneurysms.

“What machine learning has allowed us to do is move beyond simply looking at parameters. Now, we can examine the automated region of interest,” Ehlers said. “We can segment the blood vessels and focus on the areas most significantly involved. We can identify the optic nerve and the foveal center and even remove artifacts from the images.”

This approach has made the analysis more robust and reliable, though challenges with quantitative UWFA remain.

“One challenge with widefield images is that we can encounter eyelid or eyelash artifacts. The machine learning system helps identify these artifacts and remove them from the analysis,” Ehlers explained.

Detecting leakage

Advanced technology has also enabled the identification of various types of leakage, including perivascular and generalized leakage, which can help tailor treatment for individual patients by highlighting the most active disease pathways. Ehlers and his team conducted a study at Cleveland Clinic to determine how quantitative UWFA could detect leakage velocity and leakage speed.

“We examined 308 eyes from 174 subjects, and DRSS [diabetic retinopathy severity scale] grading was performed based on fundus photos,” Ehlers said. “This scale, which categorizes retinopathy as mild, moderate, severe, or proliferative, is typically used for more severe cases. There were fewer mild cases, as we had fewer angiograms available for analysis.”

Ehlers added that the study revealed a clear stepwise association between microaneurysm count, leakage index, and ischemic index.

“This suggests that quantitative UWFA may not only correlate with disease activity but could also provide a continuous biomarker, rather than the categorical methods currently used in clinical trials,” he said.

Personalized treatment

Ehlers discussed how ophthalmologists can validate these results in relation to diabetic retinopathy. He emphasized the importance of working within clinical trial protocols, such as Protocol AA, which aims to understand the role of ultra-widefield imaging. Collaborative efforts with multiple clinical trial sites through the Diabetic Retinopathy Clinical Research (DRCR) Network have demonstrated the potential for individualizing patient care.

“Through our partnership with the DRCR [Network], we were able to evaluate quantitative leakage in angiograms based on preliminary studies suggesting that leakage may be the most important biomarker for disease activity and progression risk over time,” Ehlers said. “We were also able to assess this leakage not only across the entire retina but within specific zones, such as the posterior pole, the central macula, or the retinal periphery.”

This collaboration enabled Ehlers and his team to develop strategies for treating patients with more advanced diabetic retinopathy.

“What we observed was the baseline association between initial leakage index and the severity of diabetic retinopathy,” Ehlers said. “Although we were masked to the clinical severity of the retinopathy, it was striking to see the stepwise correlation between leakage index and increasing disease severity.”

As quantitative UWFA continues to advance, Ehlers emphasizes the need for personalized treatment approaches, especially as diabetic eye diseases become more prevalent. “Quantification and pattern assessment in UWFA may open new opportunities for personalized and precision medicine,” Ehlers said.

Justis P. Ehlers, MD

E: ehlers@ccf.org

Ehlers is the Norman C. and Donna L. Harbert
Endowed Chair for Ophthalmic Research at
Cleveland Clinic Cole Eye Institute. Ehlers did not indicate any proprietary interest relevant to the subject matter.

Reference

1. Ehlers J. Quantitative ultra-widefield fluorescein angiography and diabetic retinopathy. Presented at: Fourth Annual Cleveland Eye Bank Foundation Virtual Vision Research Symposium; October 15, 2024; Cleveland, OH.