Impact of mydriasis on image gradability and automated diabetic retinopathy screening with a handheld camera. A real-world setting evaluation.

Key Points

• This research aims to investigate how mydriasis affects the quality of images for diabetic retinopathy screening and the subsequent AI analysis.
• Examined the effects of mydriasis on image quality in a real-world clinical setting.
• Analysed AI performance with and without mydriasis during diabetic retinopathy screening.
• Focused on model calibration and thresholding for non-mydriatic images.
• Significant improvement in image gradability with mydriasis.
• Enhanced AI performance in identifying diabetic retinopathy when images were mydriatic.
• Emphasized the need for optimized AI models for non-mydriatic images in resource-limited settings.