Insights into Imaging

Table 2 The evaluation of native and subtraction models; the mean and standard deviation calculated from 5 training rounds is reported for each metrics in each case

From: Generalizable attention U-Net for segmentation of fibroglandular tissue and background parenchymal enhancement in breast DCE-MRI

		r^a	DSC	Weighted DSC
Native	Breast	0.999 ± 0.001	0.950 ± 0.002	-
Native	FGT	0.985 ± 0.001	0.820 ± 0.005	0.864 ± 0.004
Subtraction	Breast	0.992 ± 0.001	0.927 ± 0.001	-
Subtraction	BPE	0.963 ± 0.004	0.628 ± 0.018	0.715 ± 0.015

^aPearson correlation coefficient between the volumes of the breast_vol, FGT_(%) (cf. Eq. (1)), BPE_(%) (cf. Eq. (2)) derived from ground truth and predicted masks. The p-values are all below 0.0001

Back to article page