Table 3.

Comparative results of U-Net with different loss functions (bold represents the best performance).

Model Name	Backbone	Batch Size	Optimizer	Loss Function	Metrics
					Accuracy	Sensitivity	Specificity	Dice	AJI
U-Net	EfficientNetB7	8	Adam	BCE+Dice	0.98894	0.82885	0.99238	0.72865	0.58277
				BCE+IoU	0.98931	0.81816	0.99305	0.73447	0.58986
				BCE+DICE+IoU	0.98893	0.81521	0.99249	0.72694	0.58024
				BCE+Focal	0.98953	0.60196	0.99745	0.66682	0.51891
				BCE+Lovasz	0.98874	0.81792	0.99226	0.72196	0.57535
				BCE+Dice+IoU+Focal	0.98916	0.81891	0.99286	0.7301	0.5845