Table 2 Qualitative comparison of segmentation performance by three evaluation metrics on 3DIRCADb
Method | BD(\(\%\)) | TD(\(\%\) | Precision(\(\%\)) | Sensitive(\(\%\)) | Dice(\(\%\))) |
|---|---|---|---|---|---|
FCN | 76.1 | 47.6 | 80.6\({\pm }\)15.3 | 73.8\({\pm }\)14.2 | 63.1\({\pm }\)15.5 |
VNet [22] | 78.6 | 60.4 | 87.6\({\pm }\)11.8 | 75.8\({\pm }\)8.4 | 65.5\({\pm }\)15.4 |
UNETR [11] | 79.7 | 74.1 | 86.1\({\pm }\)16.7 | 70.3\({\pm }\)6.6 | 66.3\({\pm }\)11.6 |
Huang et al. [12] | 80.1 | 66.1 | 97.1\({\pm }\)0.8 | 74.3\({\pm }\)10.6 | 67.5\({\pm }\)6.9 |
ResUnet [33] | 83.5 | 69.6 | 92.6\({\pm }\)1.4 | 71.9\({\pm }\)7.2 | 70.6\({\pm }\)8.5 |
Graph cuts (Sangse et al.) [27] | None | None | 74.1\({\pm }\)12 | None | None |
IBIMHAV-Net | 85.8 | 73.6 | 98.8\({\pm }\)0.3 | 78.1\({\pm }\)2.4 | 74.8\({\pm }\)9.5 |