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

Table 6

Segmentation results of all test images of the three datasets.

Image	ACC	Se	Sp	AUC
DRIVE
01_test	0.946	0.928	0.947	0.938
02_test	0.952	0.914	0.956	0.935
03_test	0.955	0.817	0.970	0.894
04_test	0.959	0.868	0.968	0.918
05_test	0.958	0.838	0.971	0.904
06_test	0.958	0.811	0.973	0.892
07_test	0.954	0.851	0.964	0.907
08_test	0.958	0.820	0.971	0.896
09_test	0.959	0.849	0.969	0.909
10_test	0.957	0.863	0.965	0.914
11_test	0.945	0.870	0.952	0.911
12_test	0.958	0.875	0.966	0.920
13_test	0.953	0.859	0.963	0.911
14_test	0.954	0.901	0.959	0.930
15_test	0.951	0.917	0.954	0.935
16_test	0.954	0.889	0.961	0.925
17_test	0.958	0.845	0.968	0.907
18_test	0.954	0.913	0.958	0.935
19_test	0.954	0.937	0.956	0.946
20_test	0.955	0.925	0.957	0.941
Avg.	0.955	0.875	0.962	0.918
STARE
im0002	0.972	0.839	0.981	0.910
im0077	0.967	0.966	0.961	0.964
im0163	0.961	0.976	0.960	0.968
im0255	0.970	0.872	0.979	0.926
im0291	0.980	0.798	0.990	0.894
Avg.	0.970	0.890	0.974	0.932
CHASE_DB1
11L	0.946	0.937	0.947	0.942
11R	0.942	0.950	0.942	0.946
12L	0.953	0.878	0.959	0.919
12R	0.958	0.872	0.965	0.918
13L	0.958	0.884	0.963	0.923
13R	0.956	0.850	0.963	0.907
14L	0.970	0.895	0.968	0.931
14R	0.966	0.867	0.971	0.919
Avg.	0.956	0.892	0.960	0.926