|
Reference | Techniques | Evaluation based on the performance |
|
Jaiswal et al. [32] | A deep transfer learning-based DenseNet201 was proposed to classify normal and COVID-19 CT images | An accuracy, sensitivity, and specificity of 96.25% |
96.62% and 96.215% were obtained |
Farooq and Hafeez [33] | A modified ResNet was developed for the classification of COVID-19 and other lung infections, namely, bacterial and viral using deep learning | An accuracy of 96.23% and a specificity of 100% were obtained |
100 |
Ahuja et al. [7] | A comparison of various deep learning models was done to classify COVID-19 and non-COVID-19 X-ray images | ResNet18 performs better compared to other transfer learning models. A classification of 99.4% is obtained for binary classification |
Apostolopoulos and Mpesiana [34] | A comparison was made on various transfer learning techniques using lung X-ray images for normal vs. COVID-19 and normal vs. bacterial vs. viral | A sensitivity, specificity, and accuracy of 98.66%, 96.46%, and 96.78% were obtained |
Ozturk et al. [35] | A deep transfer learning-based darknet was utilized for the classification of non-COVID-19, COVID-19, and pneumonia X-ray images | An accuracy of 98.08% was obtained for COVID-19 and non-COVID-19 classification, and for multiclass, an accuracy of 87.02% was obtained |
Khalifa et al. [39] | Transfer learning-based deep neural networks and GAN (generative adversarial networks) are used for the classification of COVID-19 and non-COVID-19 X-ray images | Among the considered networks, ResNet18 performed well as it gives 98.97% as F1-score |
Liang et al. [36] | A modular convolution neural network based on a multistage framework has been developed. The authors have performed a classification of the CT images | An F1-score of 98.90% and specificity of 100% were obtained |
Chakraborty et al. [37] | A new method based on deep learning called Corona-Nidaan has been used for the classification of three classes of COVID-19 chest X-rays | The authors have presented a classification accuracy of 95% and a sensitivity and specificity of 94% |
|