Table 5.

AUC, accuracy, sensitivity and specificity related to the bright-lesion detection on DIARETDB1 database for each classification method using a decision threshold $\delta =0.5$.

	Accuracy	Sensitivity	Specificity	AUC
Random Forests	0.9370 ± 0.0155	0.5096 ± 0.0478	0.9889 ± 0.0056	0.8852 ± 0.0314
Linear-SVM	0.8702 ± 0.0217	0.7396 ± 0.0528	0.8849 ± 0.0260	0.8879 ± 0.0301
RBF-SVM	0.8834 ± 0.0311	0.0.7509 ± 0.0489	0.8903 ± 0.0211	0.8901 ± 0.291
Gaussian Processes	0.8703 ± 0.0189	0.7705 ± 0.0500	0.8814 ± 0.0224	0.8971 ± 0.0286