| Guszt’av Ga’al et al. [53]
|
Attention U-Net+ adversarial+ Contrast Limited Adaptive Histogram Equalization (CLAHE) [75]
|
247 images from Japanese Society of Radiological Technology (JSRT) Dataset+ Shenzhen dataset contains a total of 662 chest X-Rays |
DSC of 97.5% on the JSRT dataset |
| Asmaa Abbas et al. [64]
|
CNN features of pre-trained models on ImageNet and ResNet+ Decompose, Transfer, and Compose (DeTraC), for the classification of COVID-19 chest X-Ray images: The developed code is available at https://github.com/asmaa4may/DeTraCCOVId19
|
I80 samples of normal CXRs (with 4020  4892 pixels) from the Japanese Society of Radiological Technology (JSRT) + Cohen JP. COVID-19 image data collection. https://githubcom/ieee8023/covid-chestxray-dataset. 2020;. |
High accuracy of 95.12% (with a sensitivity of 97.91%, a specificity of 91.87%, and a precision of 93.36%) |
| Ali Narin et al. [76]
|
Pre-trained ResNet50 model with transfer learning |
The open source GitHub repository shared by Dr. Joseph Cohen+Chest X-Ray Images (Pneumonia) https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia
|
Accuracy (97% accuracy for InceptionV3 and 87% accuracy for Inception-ResNetV2). |
| Linda Wang et al. [42]
|
COVID-Net: lightweight residual projection expansion- projection-extension (PEPX) design pattern, Model is available publicly for open access at https://github.com/lindawangg/COVID-Net. |
COVIDx dataset: 16,756 chest radiography images across 13,645 patient cases from two open access data repositories |
Accuracy 92.4% on COVIDx dataset |
| Ezz El-Din Hemdan et al. [60]
|
COVIDX-Net: based on seven different architectures of DCNNs; namely VGG19, DenseNet201, InceptionV3, ResNetV2, Inception ResNetV2, Xception, and MobileNetV2 |
COVID-19 cases provided by Dr. Joseph Cohen and Dr. Adrian Rosebrock [63]
|
F1-scores of 89% and 91% for normal and COVID-19, respectively |
| Khalid EL ASNAOUI et al. [77]
|
Fined tuned versions of (VGG16, VGG19, DenseNet201, Inception-ResNet-V2, Inception-V3, Resnet50, MobileNet-V2 and Xception). |
5856 images (4273 pneumonia and 1583 normal). |
Resnet50, MobileNet-V2 and Inception-Resnet-V2 show highly satisfactory performance with accuracy (more than 96%). |
| Prabira Kumar Sethy et al. [78]
|
Deep features from Resnet50 + SVM classification |
Data available in the repository of GitHub, Kaggle and Open-i as per their validated X-Ray images. |
Resnet50 plus SVM achieved accuracy, FPR, F1 score, MCC and Kappa are 95.38%,95.52%, 91.41% and 90.76% Respectively. |
| Ioannis D. Apostolopoulos1 et al. [79]
|
Various fine-tune dmodels: VGG19, MobileNet, Inception,Inception Resnet V2, Xception |
1427 X-Ray images. 224 images with confirmed Covid-19, 700 images with confirmed common pneumonia, and 504 images of normal conditions are included |
Accuracy with Xception was the highest, 95.57%, sensitivity of 8% and specificity of 99.99%. |
| Biraja Ghoshal et al. [58]
|
Dropweights based Bayesian Convolutional Neural Networks (BCNN) |
68 Posterior-Anterior (PA) X-Ray images of lungs with COVID-19 cases from Dr. Joseph Cohen’s Github repository, augmented the dataset with Kaggle’s Chest X-Ray Images (Pneumonia) from healthy patients, a total of 5941 PA chest radiography images across 4 classes (Normal: 1583, Bacterial Pneumonia: 2786, non-COVID-19 Viral Pneumonia: 1504, and COVID-19: 68). |
Accuracy of 89.82% with BCNN at dropweights rate=3% |
| Muhammad Farooq, Abdul Hafeez [59]
|
3-step technique to fine-tune a pre-trained ResNet-50 architecture to improve model performance |
COVIDx dataset |
Accuracy of 96.23% (on all the classes) on the COVIDx dataset |
| Yu-Huan Wu et al. [74]
|
3-class classifier (healthy, COVID-19, non-COVID-pneumonia with Res2Net backbone. Segmenation model wih VGG-16 backbone |
COVID-CS dataset (144,167 images, 750 patients of which are 400 COVID-19 positive) |
95% sensitivity and 93% specificity |
