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

Table 2

Graph-based segmentation performance without and with correction methods on DRIVE, ARIA, and STARE datasets, respectively. 2nd row: CLAHE () [3, 4, 41]. 3rd row: Retinex (, spatial spread based on low-pass filter , and geometric spread of the image intensity ) [5]. 4th row: subtraction median blurring image of green () from green image [6]. 5th row: subtraction low-pass Gaussian blurred image of green () from green image. Se: sensitivity, Sp: specificity, Acc: accuracy, and AUC: area under curve.

Datasets	Correction methods	Evaluation metrics
		Se	Sp	Acc	AUC
DRIVE	Without	0.896	0.779	0.789	0.837
	CLAHE	0.841	0.867	0.865	0.806
	Retinex	0.804	0.934	0.923	0.854
	Green-MF (Green)	0.911	0.619	0.645	0.764
	Green-GF (Green)	0.921	0.702	0.931	0.0.869
ARIA	Without	0.743	0.913	0.898	0.828
	CLAHE	0.781	0.823	0.819	0.802
	Retinex	0.768	0.701	0.707	0.734
	Green-MF (Green)	0.794	0.667	0.679	0.731
	Green-GF (Green)	0.721	0.944	0.924	0.832
STARE	Without	0.629	0.924	0.850	0.774
	CLAHE	0.643	0.902	0.838	0.770
	Retinex	0.650	0.894	0.839	0.769
	Green-MF (Green)	0.670	0.866	0.821	0.769
	Green-GF (Green)	0.563	0.975	0.879	0.774