Insights into Imaging

Table 1 Quantitative comparison of the validation and testing sets among different methods

From: Deep learning approach for automatic segmentation of ulna and radius in dual-energy X-ray imaging

Methods	Validation set (Dice)		Testing set (Dice)
Methods	Ulna	Radius	Ulna	Radius
U-Net	0.9799 ± 0.0228	0.9857 ± 0.0100	0.9804 ± 0.0208	0.9859 ± 0.0093
FCN	0.9786 ± 0.0101	0.9840 ± 0.0061	0.9787 ± 0.0100	0.9841 ± 0.0063
Ours	0.9838 ± 0.0136	0.9874 ± 0.0071	0.9835 ± 0.0142	0.9874 ± 0.0073

	Validation set (Jaccard)		Testing set (Jaccard)
	Ulna	Radius	Ulna	Radius
U-Net	0.9615 ± 0.0400	0.9720 ± 0.0187	0.9624 ± 0.0365	0.9724 ± 0.0176
FCN	0.9582 ± 0.0187	0.9685 ± 0.0115	0.9585 ± 0.0185	0.9688 ± 0.0119
Ours	0.9684 ± 0.0245	0.9752 ± 0.0135	0.9680 ± 0.0257	0.9751 ± 0.0139

The results are expressed as the mean ± standard deviation. Bold values indicate the best score obtained for ulna and radius segmentation

Back to article page