Table 3.

Quantitative segmentation using proposed against U-Net [27], Attention-UNet [28], Gated-UNet [29], Dense-UNet [30], U-Net++ [31], and Inf-Net [32] supervised methods. The best two results are shown in red and blue fonts.

Segmentation methods	Quality mertics
	Dice	Sensitivity	Specificity	Precision
U-Net [27]	0.308	0.678	0.836	0.265
Attention-UNet [44]	0.466	0.723	0.930	0.390
Gated-UNet [29]	0.447	0.674	0.956	0.375
Dense-UNet [30]	0.410	0.607	0.977	0.415
U-Net+ [31]	0.444	0.877	0.929	0.369
Inf-Net [43]	0.579	0.870	0.974	0.500
Seg-Net [33]	0.705	0.852	0.954	–
BiSe-Net [34]	0.706	0.852	0.852	–
ESP-Net [35]	0.706	0.859	0.954	–
Proposed	0.714	0.733	0.994	0.739