An Automatic Cognitive Graph-Based Segmentation for Detection of Blood Vessels in Retinal Images

Table 5

Performance of segmentation methods based on sensitivity (Se), specificity (Sp), accuracy (Acc), and area under curve (AUC) on DRIVE and STARE datasets. The hand segmented images from first-manual observers are used as benchmarks (1st STARE manual is selected because all works used it).

Methods	DRIVE			STARE
	Sensitivity	Specificity	Accuracy	Sensitivity	Specificity	Accuracy
Bankhead et al. [10]	0.703	0.028	0.9371	—	—	—
Estrada et al. [19]	—	—	0.934	—	—	—
Wang et al. [11]	—	—	0.946	—	—	0.952
Nguyen et al. [42]	—	—	—	—	—	—
Orlando and Blaschko [43]	0.785	0.967	—	—	—	—
Salazar-Gonzalez et al. [44]	0.7512	0.0316	0.941	0.789	0.037	0.944
Mithun et al. [18]	0.471	0.970	0.936	—	—	—
Azzopardi et al. [4]	0.744	0.978	0.953	0.786	0.975	0.951
Zhao et al. [5]	0.744	0.978	0.953	0.786	0.975	0.951
Imani et al. [6]	0.752	0.975	0.952	0.750	0.975	0.959
Proposed method	0.850	0.944	0.934	0.633	0.950	0.924