Table 5.

The median values of the evaluation metrics computed for the modified U-Net and InceptionV3 during the rounds of the multi-class classification task with repeated fivefold cross-validation.

CNN$\backslash$Metric	Acc.	Sen.	Spe.	Pre.		Youden	F1		$\kappa$	MCC
				Mac.	Mic.		Mac.	Mic.
U-Net	0.857	0.714	0.905	0.760	0.714	0.619	0.693	0.714	0.619	0.641
InceptionV3	0.801	0.602	0.867	0.589	0.602	0.469	0.591	0.602	0.469	0.473

The results obtained with micro-averaging (Mic.) and macro-averaging (Mac.) are here separately for precision and F1-score but both these methods give the same values for accuracy, sensitivity, specificity, and Youden’s index.