Comparative Study of Retinal Vessel Segmentation Based on Global Thresholding Techniques

Table 5

Performance of different proposed segmentation methods on STARE database.

Method	Average accuracy	Average sensitivity	Average specificity
Human observer	0.9354	0.8949	N/A
Hoover et al. [8]	0.9275	0.6751	0.9567
Staal et al. [14]	0.9516	0.6970	N/A
Jiang and Mojon [11]	0.9009	N/A	N/A
Marín et al. [19]	0.9526	N/A	N/A
Ricci and Perfetti [24]	0.9646	N/A	N/A
Soares et al. [13]	0.9480	N/A	N/A
Akram and Khan [42]	0.9502	N/A	N/A
Amin and Hong [37]	0.9081	0.7261	N/A
Mendonça and Campilho [28]	0.9479	0.7123	N/A
Tagore et al. [38]	0.9497	N/A	N/A
CLAHE and average filter (ATC) using Otsu	0.9409	0.6258	0.9662
CLAHE and Gaussian filter (ATC) using Otsu	0.9435	0.6138	0.9698
CLAHE with average and Gaussian filters (ATC) using Otsu	0.9468	0.6144	0.9735
CLAHE and adaptive filter (ATC) using Otsu	0.9456	0.6135	0.9722
CLAHE and average filter (ATC) using ISODATA	0.9421	0.6238	0.9676
CLAHE and Gaussian filter (ATC) using ISODATA	0.9442	0.6099	0.9709
CLAHE with average and Gaussian filters (ATC) using ISODATA	0.9471	0.6127	0.9740
CLAHE and adaptive filter (ATC) using ISODATA	0.9458	0.6115	0.9726
Phase congruence with IDM-based threshold (MO)	0.9340	0.5202	0.9682
Phase congruence with IDM-based threshold (MOMF)	0.9318	0.5036	0.9671
Phase congruence with IDM-based threshold (ATC)	0.9221	0.5031	0.9567