Multiscale U-Net with Spatial Positional Attention for Retinal Vessel Segmentation

Table 4

The results of different algorithms in the DRIVE dataset.

Dataset	Methods	Year	Se (%)	Ac (%)	AUC (%)	F1-score (%)
DRIVE	R2U-Net [30]	2018	77.92	95.56	97.84	81.71
	U-Net [30]	2018	75.37	95.31	97.55	81.42
	LadderNet [31]	2018	78.56	95.61	97.93	82.02
	DUNet [32]	2019	78.94	96.97	98.56	N/A
	DEU-Net [33]	2019	79.40	95.67	97.72	82.70
	AG-Net [34]	2019	81.00	96.92	98.56	N/A
	IterNet [12]	2019	77.35	95.73	98.16	82.05
	BCDU-Net [35]	2019	80.07	95.60	97.89	82.24
	Tang et al. [36]	2020	81.60	95.54	97.99	N/A
	Lü et al. [37]	2020	80.62	95.47	97.39	N/A
	SA-UNet [38]	2020	82.12	96.98	98.64	82.63
	Zhang et al. [13]	2020	81.51	96.95	98.63	N/A
	RVSeg-Net [39]	2020	81.07	96.81	98.17	N/A
	Proposed method	2021	83.16	96.99	98.76	82.91