Table 3 Classification results obtained with different deep feature sets using our proposed MRFGRO algorithm.
From: MRFGRO: a hybrid meta-heuristic feature selection method for screening COVID-19 using deep features
Feature set | SARS-CoV-2 CT-scan dataset | Covid-CT dataset | MOSMED dataset | |||
|---|---|---|---|---|---|---|
No. of selected features | Accuracy (%) | No. of selected features | Accuracy (%) | No. of selected features | Accuracy (%) | |
GoogLeNet | 780 | 94.47 | 680 | 96.22 | 811 | 91.91 |
ResNet18 | 445 | 92.17 | 328 | 96.91 | 378 | 90.11 |
ResNet152 | 1119 | 90.99 | 998 | 94.29 | 1242 | 91.49 |
VGG19 | 12,400 | 87.77 | 9442 | 85.48 | 15,987 | 81.24 |
VGG16 | 17,809 | 85.47 | 14,899 | 86.78 | 12,597 | 81.24 |
ResNet18+GoogLeNet | 875 | 99.42 | 756 | 99.15 | 612 | 95.57 |
ResNet152+GoogLeNet | 1180 | 97.71 | 987 | 96.18 | 1001 | 91.23 |
ResNet18+VGG16 | 15,489 | 90.02 | 14,801 | 92.24 | 17,589 | 92.21 |
GoogLeNet+VGG19 | 16,029 | 91.19 | 11,549 | 90.42 | 18,900 | 78.48 |
ResNet152+VGG19 | 15,014 | 88.18 | 17,802 | 85.44 | 11,259 | 80.04 |
ResNet18+GoogLeNet+VGG16 | 9002 | 86.48 | 15,809 | 84.48 | 18,792 | 79.99 |
ResNet152+GoogLeNet +VGG19 | 16,891 | 87.62 | 18,722 | 81.19 | 11,589 | 78.48 |
