Table 2 Segmentation performance of recent models on LIDC-IDRI and LNDb
From: GLANCE: continuous global-local exchange with consensus fusion for robust nodule segmentation
Model | Task | LIDC-IDRI | LNDb | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Â | Â | DSC (%) | IoU (%) | Sens. (%) | HD95 (mm) | DSC (%) | IoU (%) | Sens. (%) | HD95 (mm) |
GLANCE (ours) | Seg+Det | 95.5 | 91.4 | 96.0 | 0.7 | 95.3 | 91.0 | 95.8 | 0.8 |
UnetTransCNN19 | Seg | 92.6 | 86.2 | 93.5 | 1.2 | 90.7 | 83.0 | 91.3 | 1.4 |
BRAU-Net++25 | Seg | 93.1 | 87.1 | 93.9 | 1.1 | 91.1 | 83.7 | 91.7 | 1.3 |
CT-UFormer17 | Seg | 94.0 | 88.7 | 94.6 | 1.0 | 92.0 | 85.2 | 92.7 | 1.2 |
UNETR++19 | Seg | 93.5 | 87.8 | 94.1 | 1.1 | 91.5 | 84.3 | 92.1 | 1.3 |
Improved V-Net4 | Seg | 85.7 | 75.0 | 84.8 | 2.6 | 81.1 | 68.2 | 76.9 | 3.1 |
Alhajim et al.26 | Seg+Det | 90.9 | 83.3 | 91.4 | 1.6 | 88.7 | 79.7 | 89.3 | 1.9 |
AWEU-Net27 | Seg+Det | 90.4 | 82.5 | 90.8 | 1.8 | 87.2 | 77.3 | 86.9 | 2.2 |
Liu et al.28 | Seg | 87.5 | 77.8 | 86.8 | 2.9 | 84.1 | 72.6 | 83.4 | 3.2 |
Abdullah et al.29 | Seg | 89.8 | 81.5 | 90.4 | 2.1 | 86.9 | 76.8 | 87.6 | 2.4 |
Su et al.30 | Seg | 92.0 | 85.2 | 92.7 | 1.3 | 90.0 | 81.8 | 90.6 | 1.5 |
MCAT-Net3 | Seg | 88.3 | 79.1 | 86.3 | 3.0 | 78.5 | 64.6 | 75.1 | 3.8 |
SAtUNet6 | Seg | 81.1 | 68.2 | 80.6 | 4.1 | 79.0 | 65.3 | 78.5 | 4.3 |
STCA-Net2 | Seg | 91.6 | 84.5 | 92.1 | 1.5 | 89.8 | 81.5 | 90.4 | 1.7 |
SSLKD-UNet16 | Seg | 92.3 | 85.7 | 92.9 | 1.3 | 90.1 | 82.0 | 90.7 | 1.6 |
Niu et al.31 | Seg | 93.8 | 88.3 | 94.4 | 1.1 | 91.6 | 84.5 | 92.2 | 1.3 |
MT-Net32 | Seg+Cls | 83.2 | 71.2 | 82.6 | 3.5 | 80.1 | 66.8 | 79.3 | 3.7 |
EDTNet33 | Seg | 89.1 | 80.3 | 89.7 | 2.2 | 86.4 | 76.1 | 87.0 | 2.6 |
Wavelet U-Net++18 | Seg | 93.7 | 88.1 | 94.2 | 1.0 | 92.2 | 85.5 | 92.8 | 1.2 |
