Table 4. Comparison of liver tumor segmentation performance on the LiTS dataset of the different models.

Method	Dice	Sensitivity	ASD
U-Net	0.643	0.897	7.06
Trans U-Net	0.701	0.901	6.52
Attention U-Net	0.691	0.903	5.84
Dense U-Net	0.727	0.914	3.64
Proposed method (AGCAF-Net)	0.841	0.917	3.52

LiTS, liver tumor segmentation benchmark; ASD, average symmetric surface distance; AGCAF-Net, attention-guided context asymmetric fusion network.