Table 6. The overall and per-class performance of the GA-CNN, LCBO-CNN, MVO-CNN, SBO-CNN, WOA-CNN, and EOSA-CNN hybrid algorithms as compared with the basic CNN architecture.

Measure/ Methods	GA-CNN	LCBO-CNN	MVO-CNN	SBO-CNN	WOA-CNN	EOSA-CNN	CNN
Overall Performance
Accuracy (95% CI)	0.82	0.83	0.81	0.82	0.82	0.87	0.80
Cohens kappa	0.63	0.61	0.59	0.62	0.63	0.70	0.60
Precision	0.84	0.81	0.79	0.81	0.82	0.83	0.81
Recall	0.77	0.77	0.76	0.77	0.77	0.82	0.75
F1 score	0.80	0.78	0.77	0.79	0.79	0.82	0.78
Specificity	0.73	0.86	0.82	0.76	0.75	0.98	0.70
Sensitivity	0.57	0.45	0.33	0.43	0.41	0.38	0.53
Performance per class
Sensitivity
Normal	0.7212	0.8558	0.7885	0.7500	0.7404	0.9231	0.6827
Benign	0.5333	0.4333	0.3333	0.4333	0.400	0.36667	0.5333
Malignant	0.8643	0.7714	0.8214	0.8571	0.8786	0.8500	0.8500
Specificity
Normal	0.8941	0.7765	0.7882	0.8235	0.8588	0.8294	0.8529
Benign	0.8320	0.9016	0.8893	0.8689	0.8525	0.9508	0.8320
Malignant	0.9776	0.9851	0.9701	0.9925	0.9851	0.9478	0.9851
Precision
Normal	0.8065	0.7008	0.6949	0.7222	0.7624	0.7680	0.7396
Benign	0.2807	0.3514	0.2703	0.2889	0.2500	0.4783	0.2807
Malignant	0.9758	0.9818	0.9664	0.9917	0.984	0.9444	0.9835
Recall
Normal	0.7212	0.8558	0.7885	0.7500	0.7404	0.9231	0.6827
Benign	0.5333	0.4333	0.3333	0.4333	0.400	0.3667	0.5333
Malignant	0.8643	0.7714	0.8214	0.8571	0.8786	0.8500	0.8500
F1 score
Normal	0.7614	0.7706	0.7387	0.7358	0.7512	0.8384	0.71
Benign	0.36782	0.38806	0.2985	0.34667	0.3077	0.41509	0.36782
Malignant	0.9167	0.864	0.888	0.9195	0.9283	0.8947	0.9119
Balanced Accuracy
Normal	0.8076	0.8161	0.7883	0.7868	0.7996	0.8762	0.7678
Benign	0.68265	0.66749	0.6113	0.65109	0.6262	0.65874	0.68265
Malignant	0.9209	0.8783	0.8958	0.9248	0.9318	0.8989	0.9175