Table 2. Quantitative evaluation of vessel segmentation algorithms related to the first class of the images.

Method	TPR	FPR	ACC	Improvement ()	TPR	FPR	ACC
Our method	0.6988	0.0267	0.9504
Jiang et al. [17]	0.5993	0.031	0.9238	our method vs Jiang et al.	17	−0.14	2.9
Perez et al. [27]	0.5772	0.0367	0.9172	our method vs Perez et al.	21.1	−27.2	3.6
Staal et al. [5]	0.674	0.0178	0.9566	our method vs Staal et al.	3.7	0.5	−0.6
Zana et al. [20]	0.6287	0.0197	0.9375	our method vs Zana et al.	11.1	35.53	1.4
2nd observer	0.7825	0.0378	0.9460	our method vs 2nd observer	−10.7	−29.37	0.5

Comparison of performance between the recent studies according to the first class of the images, including 6th, 9th, 12th, 17th, 18th, 20th test images from the DRIVE database.