Table 2.

Comparation to the other four models.

Datasets	Methods	Train_ loss	Train_ acc (%)	Val_ loss	Val_ acc (%)
Four classes (X-ray images)	ResNet-50	0.1987	98.56	0.2018	93.29
	VGGNet-16	0.2145	98.14	0.2453	93.05
	i-CapsNet	0.1584	98.86	0.1862	93.25
	MGMADS-3	0.0139	99.62	0.0140	96.25
	RMT-Net	0.0132	99.64	0.0126	98.84
Binary classes (CT images)	ResNet-50	0.1454	99.01	0.1752	96.25
	VGGNet-16	0.1463	98.95	0.1568	93.75
	i-CapsNet	0.1285	98.98	0.1366	95.37
	MGMADS-3	0.0025	99.93	0.0136	98.09
	RMT-Net	0.0102	99.87	0.0114	99.24

Bold value highlights the gain effect of our method in the table.