|
Study | Features | Classifier | Best Results/Findings |
|
Hemdan [47] | Chest X-ray | VGG19, DenseNet | Accuracy: 90% (VGG19) Accuracy: 90% (DenseNet) |
Arias-Garzón et al. [48] | Chest X-ray, lung segmentation | VGG19, U-Net | Accuracy: 97.05% (VGG19) |
Bushra et al. [49] | Chest X-ray images | CNN | Accuracy: 98.65%, sensitivity: 98.49%, specificity: 98.82%, precision: 98.65%, and F1-score: 98.6% |
Sharmila and Florinabel [50] | Chest X-ray | DCGAN-CNN | Accuracy: 98.6% |
Brunese et al. [51] | Color layout descriptor | Machine learning | Precision: 0.965, recall: 0.965 |
Abugabah et al. [52] | Chest X-ray | COVID-3DS-CNN | Accuracy: 96.70%, specificity: 95.55%, and sensitivity: 96.62% |
Apostolopoulos and Mpesiana [53] | Chest X-ray | VGG19 MobileNet-V2 | Accuracy: 96.78%, sensitivity: 98.66%, and specificity: 96.46% |
Manokaran et al. [54] | Chest X-ray | DNN | Accuracy: 94% |
Madhavan et al. [55] | Chest X-ray | Res-CovNet | Accuracy: 98.4% (binary), 96.2% (multiclass) |
Rahaman et al. [56] | Chest X-ray images | VGG16, VGG19, ResNet, DenseNet, MobileNet, Xception, and Inception | Highest accuracy: 89.3% (VGG19) |
Albahli and Albattah [57] | Chest X-ray images | ResNet-V2, InceptionNet-V3, and NASNetLarge | Accuracy: 99.02% (InceptionNet) |
Gouda et al. [58] | Chest X-ray images | ResNet-50 | Accuracy: 99.63%, precision: 100%, recall: 98.89%, F1-score: 99.44, and AUC:100% |
Awan et al. [59] | Chest X-ray images | Inception-V3 ResNet-50 VGG19 | Highest accuracy: 100% (binary) |
Mahesh et al. [60] | Chest X-ray images | CNN | Accuracy: 100% |
Sarki et al. [61] | Chest X-ray images | CNN | Accuracy: 100% (binary), 93.75% (multiclass) |
Ho and Gwak [62] | Handcrafted features Radiomic features Deep features | LDA, kNN, GNB, SVM, AdaBoost, RF, ensemble XGBoost, and NN | Accuracy: 89.2%, precision: 89.2%, recall: 89.2%, and F1-score: 89.2% |
Rawat et al. [63] | Chest X-ray | Inception-V3, MobileNet, Xception, and DenseNet | Accuracy: 100% (InceptionV3) |
Zouch et al. [64] | Chest X-ray CT scan | VGG19 and ResNet-50 | Accuracy: 99.35% (VGG19), 96.77% (ResNet50) |
Aggarwal et al. [65] | Contrast-limited adaptive histogram equalization (CLAHE) | Pretrained CNNs | Accuracy: 81% |
Reshi et al. [66] | Chest X-ray | CNN | Accuracy: 99, precision: 1.0, sensitivity: 0.990, specificity: 1.0, F1-score: 0.994, and AUC: 0.990 |
Attaullah et al. [67] | Chest X-ray COVID-19 symptoms | Logistic regression and CNN | Accuracy: 78.88%, specificity: 94%, and sensitivity: 77% |
Proposed method I | Chest X-ray images | VGG16 | Accuracy: 100%, precision: 100%, recall: 100%, F1-score: 100%, G-mean: 1.0, MCC: 1.0, and AUC: 1.0 |
Proposed method II | SURF features | Machine learning algorithms | Accuracy: 98.1% linear/coarse Gaussian SVM) AUC: 0.99 (linear/course Gaussian SVM) Recall: 98.80% (linear SVM, cosine kNN) Precision: 98.70% (linear/coarse Gaussian SVM) F1-score: 98.15% (coarse Gaussian SVM) |
|