Abstract
The purpose of this study was to determine the effectiveness of image-sharpening algorithms in detecting the severity of diabetic retinopathy (DR) in ultra-widefield (UWF) fundus photographs. This was a retrospective observational study of 100 UWF fundus photographs of 100 diabetic patients. The Optos UWF photographs of eyes with DR were enhanced using medical image enhancement software, and 3 masked retinal specialists evaluated the severity of DR according to the International Clinical Diabetic Retinopathy (ICDR) Disease Severity Scale. This scale is a five-stage classification system for DR. The skewness and kurtosis were measured to assess the image contrast. The results showed that the severity of the DR in the original and enhanced images did not agree in 14 of the 100 eyes. Eleven eyes were classified to have more severe retinopathy in the enhanced images than in the regular images. Twenty intraretinal retinal hemorrhages were detected in each of the four quadrants, definite venous beading in two quadrants, and prominent intraretinal microvascular abnormalities. These lesions were detected significantly more often in the enhanced images (P = 0.03). The skewness and kurtosis were significantly improved in the enhanced images (P < 0.001, P < 0.001, respectively). We conclude that the image-sharpening algorithms can improve the detection of lesions in the Optos UWF images, and improve the accuracy of the severity index in DR. These improvements make the image-sharpening algorithm a valuable tool for assessing the severity of DR.