SERR-U-Net: Squeeze-and-Excitation Residual and Recurrent Block-Based U-Net for Automatic Vessel Segmentation in Retinal Image

Table 9

Comparative results with state-of-the-art methods on STARE databases.

STARE	Methods	ACC	SE	SP	AUC
Unsupervised learning	Lam [29]	0.9567	\	\	0.9739
	You [30]	0.9497	0.7260	0.9756	\
	Azzopardi [31]	0.9563	0.7716	0.9701	0.9497
Supervised learning	Roychowdhury [32]	0.9510	0.7720	0.9730	0.9690
	Liskowsk [33]	0.9566	0.7867	0.9754	0.9785
	Qiaoliang [34]	0.9628	0.7726	0.9844	0.9879
	Ours	0.9796	0.8220	0.9926	0.9859