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

Table 4

Comparison of segmentation effects of different loss functions in VNet and WGAN fusion models (ours).

Loss function	Accuracy	Dice
IoU loss	0.93	0.86
Cross-entropy loss	0.95	0.89
Dice loss	0.94	0.91
Boundary loss	0.92	0.88
Dice loss+boundary loss (ours)	0.94	0.92

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