Deep Ensemble Learning to Detect Retinal Vascular Leakage on Ultrawide-Field Fundus Photographs of Patients With Uveitis

Puntos clave

• The aim is to develop an algorithm to detect retinal vascular leakage on ultrawide-field images.
• Utilized deep ensemble learning techniques
• Assessed images from patients with uveitis
• Evaluated detection performance against standard screening methods
• Algorithm accurately identifies retinal vascular leakage in ultrawide-field images
• Shows potential as a screening tool in resource-limited settings
• May reduce the need for fluorescein angiography in certain patients