TABLE 3.

Comparison of segmentation results between various methods (Optimal value for each evaluation metric is shown in bold. In the top row of this and subsequent tables, up arrows indicate that an increase in the evaluation metric implies better performance, the reverse being the case for the down arrows).

Method	DSC↑	Recall↑	Precision↑	ASSD↓	HD↓
3D-UNet (Çiçek et al., 2016)	0.837	0.844	0.837	0.613	30.076
VNet (Milletari et al., 2016)	0.837	0.810	0.872	0.538	33.519
ResUNet (Lee et al., 2017)	0.841	0.832	0.882	0.533	29.054
DenseUNet (Li et al., 2018)	0.839	0.826	0.859	0.514	37.413
AttUNet (Islam et al., 2020)	0.843	0.835	0.857	0.506	32.642
UNETR (Hatamizadeh et al., 2022)	0.827	0.784	0.884	0.551	44.071
UCTransNet (Wang et al., 2022a)	0.818	0.821	0.813	1.205	59.128
DR-LCT-UNet (Ours)	0.858	0.863	0.858	0.425	28.215