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

Table 5

Comparison of the proposed method with other methods on the CHASE_DB1 dataset.

Method	Se	Sp	ACC	AUC
Biswal et al. (2017) [30]	0.76	0.97	—	0.865
Yan et al. (2018) [13]	0.7633	0.9809	0.9610	0.8721
Oliveira et al. (2018) [15]	0.7779	0.9864	0.9653	0.8822
Wang et al. (2019) [25]	0.7730	0.9792	0.9603	0.8761
Guo et al. (2019) [31]	0.7888	0.9801	0.9627	0.8845
Soomro et al. (2019) [39]	0.8020	0.968	0.891	0.885
Tamim et al. (2020) [38]	0.7585	0.9846	0.9577	0.8716
Joshua et al. (2020) [40]	0.7796	0.9864	0.9722	0.8830
2^nd human observer	0.7686	0.9779	0.9560	0.8733
Proposed method	0.8916	0.9596	0.9561	0.9256