Compound W-Net with Fully Accumulative Residual Connections for Liver Segmentation Using CT Images

Table 3

Quantitative comparison between our models and other models using 3Dircadb dataset.

Model	Dice	Testing
	%	Normal data Dice %	Augmented data Dice %
U-Net [16]		46.49	60.07
Bridge U-Net [17]		73.21	60.31
Our modified bridge		79.12	60.10
Our compound model		89.88	94.42
U-Net [32]
FCN-8 s [32]
3D DSN [32]
DecNet [32]
FCN [32]	93.30
Cascaded UNet+CRF [32]	93.10
DC U-Net [32]
mU-Net [20]
UNET [28]	72.90
Cascaded UNET [28]	93.10
Cascaded UNET +3D CRF [28]	94.30