Table 1.

Related works on COVID-19 segmentation approache.

Literature	Segmentation approach	CT-IM	Patients	Performance
Zhou et al. [17]	U-Net and Focal Tversky	473	–	Dice = 83.1%, SEN = 86.7%, SPE = 99.3%
Cao et al. [18]	U-Net architecture	–	2	–
Gozes et al. [19]	Deep Learning CT Image	–	157	AUC = 0.99, SPE = 92.2%
Qiu et al. [20]	MiniSeg	100	–	SPE = 97.4%, Dice = 77.2%
Jin et al. [21]	UNet+, CNN	–	723	AUC = 0.99, SEN = 0.97, SPE = 0.92
Fei Shan et al. [22]	VB-Net	249	249	Dice = 91.6%
Xie et al. [23]	Contextual two stage U-Net	204	–	IOU = 0.91, HD95 = 6.44
Shen et al. [24]	Region growing	–	44	R = 0.7679, P < 0.05
Yan et al. [25]	COVID-SegNet	861	–	Dice = 72.60%, SEN = 75.10%.
Jun et al. [26]	UNet++	46,096	106	SPE = 93.5%, ACC = 95.2%
Zheng et al. [27]	Pretrained UNet	–	540	ROC-AUC = 0.959
Xiaowei et al. [28]	VNET	618	106	ACC = 86.7%
Fan et al. [29]	Semi-Inf-Net	100	–	Dice = 73.9%, SEN = 72.5%
Cheng et al. [30]	2D-CNN	970	496	ACC = 94.9%, AUC = 97.9%
Chen et al. [31]	U-Net Res	110	–	Dice = 94.0%, ACC = 89.0%
Ophir et al. [19]	U-net architecture	–	56	AUC = 0.99, SEN = 98.2%, SPE = 92.2%
Lin Li [32]	U-Net	4356	3,322	SEN = 90%, SPE = 96%