Table 6.

Accuracies of Vess-Net and existing methods for CHASE-DB1 dataset (unit: %).

Type	Method	Se	Sp	AUC	Acc
Handcrafted local feature-based methods	Fraz et al. [24]	72.2	74.1	-	94.6
	Pandey et al. [27]	81.06	95.30	96.33	94.94
	Sundaram et al. [29]	71.0	96.0	-	95.0
Learned/deep feature-based methods	Zhang et al. (without post-processing) [37]	77.86	96.94	-	94.97
	Zhang et al. (with post-processing) [37]	76.44	97.16	-	95.02
	Wang et al. [40]	77.30	97.92	-	96.03
	Fu et al. [45]	71.30	-	-	94.89
	Yan et al. [50]	76.41	98.06	97.76	96.07
	Jin et al. [52]	81.55	97.52	98.04	96.10
	Leopold et al. [53]	86.18	89.61	87.90	89.36
	Vess-Net (this work)	82.06	98.41	98.0	97.26