Multichannel Retinal Blood Vessel Segmentation Based on the Combination of Matched Filter and U-Net Network

Table 4

Comparison of the proposed method with other methods on the STARE dataset.

Method	Se	Sp	ACC	AUC
Khan et al. (2016) [9]	0.7359	0.9708	0.9502	0.8534
Khan et al. (2016) [35]	0.7728	0.9649	0.9518	0.8689
Khan et al. (2017) [36]	0.778	0.966	0.951	0.872
Soomro et al. (2017) [11]	0.748	0.922	0.948	0.835
Biswal et al. (2017) [30]	0.70	0.97	0.95	0.835
BahadarKhan et al. (2017) [37]	0.758	0.963	0.946	0.861
Yan et al. (2018) [13]	0.7581	0.9846	0.9612	0.8714
Oliveira et al. (2018) [15]	0.8315	0.9858	0.9694	0.9087
Wang et al. (2019) [25]	0.7523	0.9885	0.9640	0.8704
Guo et al. (2019) [31]	0.8201	0.9828	0.9660	0.9015
Feng et al. (2019) [14]	0.7709	0.9848	0.9633	0.8779
Dharmawan et al. (2019) [18]	0.7924	0.9827	—	0.8876
Saroj et al. (2020) [5]	0.7278	0.9724	0.9509	0.8501
Tamim et al. (2020) [38]	0.7806	0.9825	0.9632	0.8816
2^nd human observer	0.8952	0.9384	0.9349	0.9168
Proposed method	0.8903	0.9744	0.9699	0.9323