Automatic segmentation model of intercondylar fossa based on deep learning: a novel and effective assessment method for the notch volume

Table 1 Average Results for 5-Fold Cross-Validation with Different Networks

Network	Dice similarity coefficient	Automatic Segmentation Volume (cm³)	Manual Segmentation Volume (cm³)	Relative Error
U-Net	0.914 ± 0.04	6.483 ± 1.500	6.874 ± 1.644	0.061 ± 0.037
Seg-Net	0.906 ± 0.10	6.384 ± 1.484	6.874 ± 1.644	0.072 ± 0.036
Res-UNet	0.916 ± 0.04	6.576 ± 1.492	6.874 ± 1.644	0.047 ± 0.036
Dense-UNet	0.901 ± 0.08	6.347 ± 1.474	6.874 ± 1.644	0.077 ± 0.035
Mobile-UNet	0.906 ± 0.07	6.312 ± 1.452	6.874 ± 1.644	0.085 ± 0.051

ISSN: 1471-2474