Table 3.

Experimental performance benchmarking (mean ± standard deviation).

Model	Accuracy	Sensitivity	Specificity	F1-Score
2D ResNet-50	0.620 ± 0.052	0.604 ± 0.0078	0.636 ± 0.061	0.613 ± 0.058
3D DenseNet + H	0.713 ± 0.047	0.711 ± 0.041	0.716 ± 0.064	0.713 ± 0.043
3D DenseNet + SepConv + H	0.733 ± 0.049	0.729 ± 0.069	0.738 ± 0.047	0.731 ± 0.054
3D DenseNet	0.640 ± 0.034	0.636 ± 0.034	0.644 ± 0.048	0.638 ± 0.032
3D DenseNet + NLB + SepConv	0.878 ± 0.039	0.871 ± 0.054	0.884 ± 0.075	0.877 ± 0.041
Proposed	0.922 ± 0.037	0.929 ± 0.053	0.916 ± 0.047	0.922 ± 0.038
2D ResNet-101	0.829 ± 0.043	0.822 ± 0.054	0.836 ± 0.061	0.828 ± 0.043
3D DenseNet-201	0.856 ± 0.033	0.871 ± 0.047	0.840 ± 0.055	0.858 ± 0.032
2D VGG-19	0.667 ± 0.033	0.662 ± 0.069	0.671 ± 0.059	0.664 ± 0.041
Previous [34]	0.861 ± 0.023	0.815 ± 0.011	0.883 ± 0.048	0.810 ± 0.037