Retinal Vessel Segmentation by Deep Residual Learning with Wide Activation

Table 3

Comparison of the performance among different methods on the STARE dataset.

Type	Method	Year	F1	Sens	Spec	Acc	AUC
Unsupervised	Roychowdhury et al. [37]	2015	——	0.7317	0.9842	0.9560	0.9673
	Azzopardi et al. [38]	2015	——	0.7716	0.9701	0.9497	0.9563
	Zhang et al. [39]	2016	——	0.7791	0.9758	0.9554	0.9748
Supervised	Li et al. [40]	2015	——	0.7726	0.9844	0.9628	0.9879
	Liskowski and Krawiec[17]	2016	——	0.7867	0.9754	0.9566	0.9785
	Roychowdhury et al. [44]	2016	——	0.7720	0.9730	0.9510	0.9690
	Chen [41]	2017	——	0.7295	0.9696	0.9449	0.9557
	Yan et al. [45]	2018	——	0.7581	0.9846	0.9612	0.9801
	Dense U-Net [36]	2019	——	0.7914	0.9722	0.9538	0.9704
	WA-Net (ours)	2020	0.8223	0.7740	0.9871	0.9645	0.9825