Table 2.

Performance Metrics of Fine-tuned second-stage pre-trained models for COVID-19 detection.

Models	Technique	Acc	S	SP	P	F1	MCC	K	DOR	AUC
IRCNN	Baseline	0.847	0.833	0.867	0.842	0.847	0.694	0.694	30.79	0.928 (0.886, 0.970)
	Fine-tuned	0.854	0.902	0.888	0.884	0.865	0.736	0.736	44.4	0.917 (0.872, 0.962)
Mobile Net	Baseline	0.854	0.902	0.875	0.839	0.855	0.698	0.666	35.31	0.932 (0.891, 0.95
	Fine-tuned	0.875	0.902	0.819	0.833	0.866	0.724	0.7222	42.17	0.904 (0.856, 0.952)
FITNET	Baseline	0.868	0.847	0.888	0.847	0.844	0.736	0.736	44.4	0.921 (0.877, 0.965)
	Fine-tuned	0.875	0.902	0.902	0.895	0.863	0.737	0.736	46.47	0.930 (0.888, 0.971)
ResNet-18	Baseline	0.833	0.916	0.847	0.884	0.865	0.714	0.708	41.83	0.930 (0.888, 0.971)
	Fine-tuned	0.895	0.861	0.902	0.897	0.878	0.751	0.752	51.54	0.981 (0.864, 0.957)
Efficient Net	Baseline	0.847	0.847	0.791	0.814	0.862	0.673	0.694	30.06	0.915 (0.868, 0.960)
	Fine-tuned	0.868	0.847	0.930	0.930	0.892	0.793	0.791	83.2	0.947 (0.913, 0.985)