A Liver Segmentation Method Based on the Fusion of VNet and WGAN

Table 2

The experimental results of different methods were compared.

Comparison method	Dataset	Accuracy	Dice
3D UNet	LiTS	0.98	0.91
3D UNet	CHAOS	0.96	0.89
Improved 2.5D VNet	LiTS	0.96	0.89
3D VNet	LiTS	0.95	0.90
2.5D UNet	LiTS	—	0.77
Attention [34]	LiTS	—	0.76
CycleGAN [35]	LiTS	0.86	0.89
DenseNet	LiTS	—	0.91
Ours (VNet_WGAN)	LiTS	0.94	0.92
Ours (VNet_WGAN)	CHAOS	0.94	0.90

Note: “-” means no data.