¹⁸F-FDG-PET/CT Whole-Body Imaging Lung Tumor Diagnostic Model: An Ensemble E-ResNet-NRC with Divided Sample Space

Table 3

Comparison of accuracy, standard deviation, and training times in different models.

	Evaluation index	AlexNet+NNC	AlexNet+Softmax	AlexNet+NRC	ResNet-50+NNC	ResNet-50+Softmax	ResNet-50+NRC
CT	Acc (%)	96.37	98.20	98.80	97.00	98.13	99.07
	SD (%)	1.29	1.14	0.93	1.51	1.34	0.50
	Training time (s)	161.48	164.33	185.91	1176.91	1182.28	1204.95
PET	Acc (%)	99.57	99.50	99.83	99.50	99.63	99.80
	SD (%)	0.75	1.31	0.25	1.31	1.30	0.48
	Training time (s)	146.63	148.77	169.46	1035.85	1036.29	1059.48
PET/CT	Acc (%)	96.60	97.63	97.97	97.53	97.90	98.33
	SD (%)	5.04	3.60	3.37	3.14	2.63	3.01
	Training time (s)	139.25	138.29	162.38	1162.93	1169.76	1187.22
Ensemble	Acc (%)	99.10	98.87	99.43	99.23	99.33	99.57
	SD (%)	2.03	2.28	0.86	1.67	1.45	1.07
	Training time (s)	422.39	420.09	491.20	3080.76	3234.90	3179.38