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

Table 3

Graph-based segmentation performance of four enhancement methods: local phase-based [5], wavelet-based [10], multiscale multiorientation BTH, and multiscale BTH methods on DRIVE, ARIA, and STARE datasets, respectively. Se: sensitivity, Sp: specificity, Acc: accuracy, and AUC: area under curve.

Datasets	Enhancement methods	Evaluation metrics
		Se	Sp	Acc	AUC
DRIVE	Phase-based	0.820	0.827	0.826	0.823
	Wavelet-based	0.827	0.768	0.773	0.798
	Multiscale multiorientation BTH	0.635	0.924	0.945	0.803
	Multiscale BTH	0.863	0.930	0.924	0.893
ARIA	Phase-based	0.469	0.818	0.785	0.644
	Wavelet-based	0.725	0.683	0.687	0.704
	Multiscale multiorientation BTH	0.813	0.893	0.871	0.828
	Multiscale BTH	0.746	0.947	0.920	0.841
STARE	Phase-based	0.432	0.776	0.855	0.699
	Wavelet-based	0.469	0.773	0.711	0.620
	Multiscale multiorientation BTH	0.660	0.944	0.888	0.803
	Multiscale BTH	0.606	0.977	0.910	0.800