A NEW TECHNIQUE USING ULTRA-WIDEFIELD CAMERA FOR IMPROVED NEOVASCULARIZATION DETECTION IN DIABETIC RETINOPATHY.

PURPOSE: To evaluate whether vascular enhancement processing of ultra-widefield fundus camera (UWFC) images improves the diagnostic performance for proliferative diabetic retinopathy (PDR), including sensitivity, specificity, and neovascularization (NV) detection accuracy.

METHODS

This retrospective cross-sectional study included 30 eyes of 24 patients with diabetic retinopathy (DR) who underwent UWFC imaging at Kagoshima University Hospital. Two image sets were prepared: original UWFC color images and UWFC images combined with vascular enhancement images generated from green-filtered overlay images using automated filtering techniques, including Gaussian, top-hat, homomorphic filtering, and unsharp masking. Ten ophthalmologists independently assessed both sets to determine NV presence and diagnose PDR. The main outcomes were sensitivity and specificity for PDR diagnosis and accuracy of NV detection, compared between sets using the Wilcoxon signed-rank test ( P < 0.05).

RESULTS

Sensitivity for diagnosing PDR significantly improved from 83.5% ± 9.7% with UWFC images alone to 90.9% ± 8.4% with the combined set ( P = 0.033). Specificity showed no significant difference (70.1% ± 15.3% vs. 66.6% ± 28.2%, P = 0.662). NV detection accuracy improved significantly from 37.9% ± 8.2% to 52.8% ± 9.7% ( P < 0.01).

CONCLUSION

Combining vascular enhancement with UWFC images significantly improves sensitivity for PDR and NV detection accuracy while maintaining specificity, providing a simple, noninvasive, and practical tool for DR screening.