Table 4.

Results of multimodal fusion.

CNN model	Class no.	Acc	AUC	Spec	Sen	F1	Kappa	Avg.Acc	Avg.F1	Avg.AUC
SqueezeNet1_1	Class 0	0.948	0.965	1.0	0.930	0.909	0.873	0.896	0.895	0.931
	Class 1	0.926	0.866	0.743	0.990	0.839	0.792
	Class 2	0.948	0.955	0.969	0.942	0.896	0.816
	Class 3	0.970	0.939	0.879	1.0	0.935	0.916
VGG 19	Class 0	0.956	0.970	1.0	0.940	0.921	0.890	0.911	0.910	0.956
	Class 1	0.948	0.900	0.800	1.0	0.889	0.856
	Class 2	0.956	0.971	0.942	1.0	0.914	0.885
	Class 3	0.963	0.924	0.848	1.0	0.918	0.894
ResNet 50	Class 0	0.971	0.980	0.900	1.0	0.947	0.927	0.918	0.919	0.953
	Class 1	0.934	0.887	0.923	0.936	0.842	0.801
	Class 2	0.934	0.963	0.848	0.978	0.897	0.848
	Class 3	0.971	0.929	1.0	0.964	0.923	0.857
DenseNet 121	Class 0	0.971	0.980	0.900	1.0	0.947	0.928	0.918	0.920	0.939
	Class 1	0.929	0.871	0.920	0.930	0.821	0.777
	Class 2	0.907	0.963	0.848	0.936	0.857	0.788
	Class 3	0.950	0.946	0.862	0.973	0.877	0.846