TABLE 3.
State‐of‐the‐art comparison
| Study | Year | Types of images used | Method(s) | Accuracy |
|---|---|---|---|---|
| Rahimzadeh et al. 14 | 2021 | CT scan | ResNet50V2 | 98.49% |
| Rahimzadeh and Attar 28 | 2020 | x‐ray images | ResNet50V2 | 91.4% |
| Md. Zabirul Islam et al. 4 | 2020 | x‐ray images | CNN‐LSTM | 99.4% |
| Rohilaa et al. 29 | 2021 | x‐ray images | ReCOV‐101 | 94.9% |
| Polat et al. 30 | 2020 | Chest CT | CNN | 93.26% |
| Modzelewski et al. 31 | 2020 | Chest CT | Deep learning‐based multi‐task model | 86% |
| Ying et al. 32 | 2020 | Chest CT | DRE‐Net | 86% |
| Butt et al. 33 | 2020 | Chest CT | ResNet with location attention | 86.7% |
| Ozturk et al. 34 | 2020 | Chest x‐ray | Dark Covid Net‐19 | 87.02% |
| Li et al. 35 | 2020 | Chest x‐ray | DenseNet | 88.9% |
| Li et al. 35 | 2020 | Chest x‐ray | COVIDX‐Net | 90% |
| Wang and Wong 36 | 2020 | Chest x‐ray | Tailored deep CNN | 92.6% |
| Proposed method | x‐ray, CT scan, MRI | CNN, DNN | 94.6% |