Skip to main content
. 2022 May 12;12:7868. doi: 10.1038/s41598-022-11852-y

Table 4.

Performance comparison on clinical tongue image segmentation (mean ± standard deviation).

Network Accuracy Sensitivity Dice AUC BF-Score
U-Net1 0.9985 ± 0.0024 0.8836 ± 0.2339 0.9025 ± 0.2010 0.7913 ± 0.1169 0.8969 ± 0.1799
CE-Net14 0.9987 ± 0.0011 0.9356 ± 0.1711 0.9231 ± 0.1681 0.8823 ± 0.0855 0.9372 ± 0.0820
MutiResUNet7 0.9984 ± 0.0022 0.9147 ± 0.1825 0.9183 ± 0.1386 0.8818 ± 0.0912 0.8893 ± 0.1593
Attention U-Net16 0.9983 ± 0.0029 0.8791 ± 0.2533 0.8862 ± 0.2170 0.8773 ± 0.1266 0.8603 ± 0.2204
ResNet5013 0.9990 ± 0.0005 0.9417 ± 0.0644 0.9547 ± 0.0375 0.8659 ± 0.0322 0.9260 ± 0.1028
nnUnet43 0.9993 ± 0.0005 0.9687 ± 0.0275 0.9678 ± 0.0198 0.9833 ± 0.0151 0.8783 ± 0.1554
MEA-Net (ours) 0.9993 ± 0.0004 0.9701 ± 0.0208 0.9704 ± 0.0141 0.9849 ± 0.0104 0.9521 ± 0.0657

Significant values are in bold.