Table 4. Conventional and proposed U-Net best performing training and validation results with various evaluation metrics.

Model	Loss	Dice	IoU	Recall	Precision	Accuracy
U-Net training	−0.9842	0.9842	0.9699	0.9808	0.9869	0.9918
U-Net validation	−0.9398	0.9398	0.8977	0.9422	0.9576	0.9749
Proposed U-Net training	−0.9708	0.9708	0.9449	0.9658	0.9760	0.9853
Proposed U-Net validation	−0.9605	0.9607	0.9265	0.9525	0.9694	0.9800