Table 4.

AUC, accuracy, sensitivity and specificity related to the bright-lesion detection on E-OPHTHA exudates database. Results were optimized to the best sensitivity/specificity trade-off.

	Accuracy	Sensitivity	Specificity	$\mathit{\delta }$
Random Forests	0.8410 ± 0.0181	0.8418 ± 0.0178	0.8411 ± 0.0181	0.8999 ± 0.0214
Linear-SVM	0.8242 ± 0.0279	0.8243 ± 0.0284	0.8242 ± 0.0278	0.5771 ± 0.0958
RBF-SVM	0.8529 ± 0.0190	0.8531 ± 0.0193	0.8529 ± 0.0190	0.5835 ± 0.1012
Gaussian Processes	0.8581 ± 0.0221	0.8579 ± 0.0221	0.8579 ± 0.0222	0.5348 ± 0.0747