Skip to main content
. 2021 Jun 12;183:115401. doi: 10.1016/j.eswa.2021.115401

Table 5.

Performance metrics for the five models trained with synthetic COVID-19 image.

Models Class TPR TNR FPR FNR PPV NPV PL NL DOR
VGG16 COVID-19
Pneumonia
Normal
99.2%
98.8%
96.9%
99.7%
98.4%
99.5%
0.3%
1.6%
0.5%
0.8%
1.2%
3.1%
99.4%
98.4%
97.7%
0.6%
1.6%
2.3%
330.73
63.75
206.21
0.0078
0.0119
0.0309
42401.3
5357.1
6673.5
MobileNetV2 COVID-19
Pneumonia
Normal
97.3%
86.9%
72.3%
90.4%
90.5%
98.9%
9.6%
9.5%
1.1%
2.7%
13.1%
27.7%
83.7%
89.9%
93%
16.3%
10.1%
7%
10.15
9.14
65.72
0.0298
0.1446
0.2800
340.6
63.2
234.7
Xception COVID-19
Pneumonia
Normal
99.4%
98.4%
94.6%
99.8%
97.8%
99.2%
0.2%
2.2%
0.8%
0.6%
1.6%
5.4%
99.6%
97.8%
96%
0.4%
2.2%
4%
497.05
44.94
121.29
0.0058
0.0118
0.0544
85698.3
3808.5
2229.6
NASNetmobile COVID-19
Pneumonia
Normal
90.4%
98.3%
81.3%
100%
87.6%
98.9%
0%
12.4%
1.1%
9.6%
1.7%
18.7%
100%
88.8%
93.4%
0%
11.2%
6.6%

7.90
73.97
0.096
0.0195
0.1883

405.1
392.8
InceptionResNet
V2
COVID-19
Pneumonia
Normal
99.2%
96.3%
95%
99.2%
97.9%
98.3%
0.8%
2.1%
1.7%
0.8%
3.7%
5%
98.5%
97.9%
92.2%
1.5%
2.1%
7.8%
125.59
46.76
55.23
0.0078
0.0374
0.0508
16101.3
1250.3
1087.2