Table 1.

Overview of studies using deep learning approaches with their performance for COVID-19 case detection.

Reference	Number of Image Samples	Classes	Architectures	Best Performing Architecture	Performance/Accuracy
[35]	1428	3	VGG19, MobileNetV2, Inception, Xception, InceptionResNetV2	MobileNetV2	Acc = 96.78%
[36]	204	2	VGG16 + Resnet50	VGG16 + Resnet50 + custom CNN	Acc = 89.2%
[37]	100	2	ResNet50, InceptionV3 and InceptionRes-NetV2	ResNet50	Acc = 98%
[38]	21,152	2	CNN	CNN	Acc = 94.64%
[39]	5184	2	ResNet18, ResNet50, SqueezeNet, DenseNet-121	SqueezeNet	Sensitivity = 98%, Specificity = 92.9%
[40]	13,975	2	COVID-CAPS	COVID-CAPS	Acc = 95.7%,
[41]	400	2	VGG16, InceptionResNetV2, ResNet50, DenseNet201, VGG19, MobilenetV2, NasNetMobile, and ResNet15V2	NasNetMobile	Acc = 93.94%
[42]	75	2	VGG19, Xception, ResNetV2, DenseNet201, InceptionV3, MobileNetV2, InceptionResNetV2	VGG19, DenseNet	F1 scores = 0.91
[43]	1127	2	Modified Darknet	Modified Darknet	Acc = 98%
[44]	1257	3	Xception	Xception	Acc = 94%
[45]	2356	3	ACoS system	ACoS	Acc = 91.33%
[46]	6100	3	SVM, LR, nB, DT, and kNN + VGG16, ResNet50, MobileNetV2, DenseNet121	Mean result	Acc = 98.5%
[47]	1428	2	VGG16	VGG16	Acc = 96%
[48]	79,500	3	Grad-CAM	Grad-CAM	Acc = 91.5%
[49]	6200	4	CSEN-based classifier	CSEN-based Classifier	Sensitivity = 98% Specificity = 95%
[50]	13,975	19	COVID-Net	COVID-Net	Acc = 93.3%
[51]	196	3	DeTrac	Detrac	Acc = 93.1%
[52]	3150	3	CapsNet	CapsNet	Acc = 97%
[53]	1127	3	Xception	Xception	Acc = 97%
[54]	7470	2	MD-Conv	MD-Conv	Acc = 93.4%
[55]	380	2	Novel CNN Model	Novel CNN Model	Acc = 91.6%
[56]	247	2	BMO-CRNN	BMO-CRNN	Sensivity = 97.01% Acc = 97.31% F-value = 97.53%