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

Table 3

Comparison of segmentation effects of different loss functions in the 2.5D VNet model.

Loss function	Accuracy	Dice
IoU loss	0.96	0.80
Cross-entropy loss	0.95	0.84
Dice loss	0.96	0.88
Boundary loss	0.97	0.83
Dice loss+boundary loss (ours)	0.96	0.89

Note: the bold font is the optimal value for each column.