Table 8.

The results of different models on the testing set in the RS dataset + NG dataset during the 10-fold cross-validation (mean ± standard deviation).

Models	Evaluation Metrics			Training Time (mins:secs)
	Acc	Se	Sp
VGG-16	0.949 ± 0.009	0.958 ± 0.009	0.931 ± 0.010	535:25
VGG-19	0.928 ± 0.013	0.933 ± 0.012	±0.912 ± 0.013	243:14
ResNet-18	0.902 ± 0.011	0.911 ± 0.010	0.901 ± 0.011	361:12
ResNet-50	0.925 ± 0.012	0.934 ± 0.011	0.914 ± 0.011	302:57
DenseNet-121	0.887 ± 0.011	0.896 ± 0.012	0.867 ± 0.012	332:13
AlexNet	0.919 ± 0.009	0.930 ± 0.009	0.898 ± 0.010	435:79
Proposed CNN	0.955 ± 0.005	0.966 ±0.007	0.947 ± 0.008	179:26

RS dataset + NG dataset = 22,062 images in the LVDD group and 28,818 images in the control group.