A Novel Method for COVID-19 Detection Based on DCNNs and Hierarchical Structure

Table 3

Results using different classification models for COVID-19 detection.

Method	Performance metrics
	Sensitivity	Specificity	Precision	-score	Accuracy
DenseNet-121 [19]	0.9159	0.9200	0.9227	0.9193	0.9200
Chen et al. [30]	0.9693	0.9700	0.9707	0.9701	0.9700
VGG-16 [14]	0.8724	0.8900	0.9133	0.8924	0.8900
Xception [35]	0.6800	0.6800	0.9273	0.7846	0.6800
Apostolopoulos et al. [31]	0.9487	0.9533	0.9604	0.9545	0.9533
Gayathri et al. [22]	0.9754	0.9402	0.9435	0.9596	0.9583
Bargshady et al. [13]	0.9001	0.8755	0.8877	0.8990	0.8769
Irfan et al. [32]	0.8824	0.9222	0.5945	0.7112	0.9220
Almalki et al. [33]	0.9628	0.9621	0.9628	0.9628	0.9625
Nguyen et al. [34]	0.9628	0.9633	0.9638	0.9633	0.9633
DualCheXNet [23]	0.8051	0.8100	0.9959	0.8904	0.8100
The proposed method	0.9900	1.0000	1.0000	0.9950	0.9967