Skip to main content
. 2021 Aug 6;136:104744. doi: 10.1016/j.compbiomed.2021.104744

Table 8.

Results using features extracted with pre-trained CNNs and an MLP classifier.

Dataset A R P F K AUROC AUPRC
VGG16
SARS-COV-2
CT-Scan
0.959 0.959 0.959 0.959 0.919 0,959 0,959
COVID-CT 0.877 0.876 0.874 0.875 0.750 0.876 0.874
VGG19
SARS-COV-2
CT-Scan
0.965 0.965 0.966 0.966 0.932 0.966 0.965
COVID-CT 0.863 0.862 0.863 0.862 0.725 0.862 0.861
Xception
SARS-COV-2
CT-Scan
0.957 0.957 0.957 0.957 0.915 0.957 0.957
COVID-CT 0.870 0.870 0.871 0.870 0.741 0.870 0.870
ResNet50
SARS-COV-2
CT-Scan
0.975 0.976 0.975 0.975 0.952 0.975 0.975
COVID-CT 0.823 0.820 0.822 0.821 0.642 0.820 0.821
Inception-v4
SARS-COV-2
CT-Scan
0.949 0.950 0.950 0.949 0.900 0.950 0.949
COVID-CT 0.836 0.833 0.839 0.835 0.670 0.833 0.833

Values in bold indicate the best results for each architecture.