Table 1.

Segmentation performance measured as Dice score for kidney, liver and spleen. Shown are mean scores and standard deviation of 5 cross-validation experiments.

Organ	Evaluation Dataset*	Augmentation Method
		None	Standard	Histogram Eq	CycleGAN
Kidney	in-distribution (contrast CT)	0.920 ± 0.013	0.940 ± 0.007	0.939 ± 0.006	0.944 ± 0.009
	out-of-distribution (non-contrast CT)	0.059 ± 0.034	0.090 ± 0.039	0.066 ± 0.027	0.664 ± 0.040
Liver	in-distribution (contrast CT)	0.944 ± 0.005	0.941 ± 0.006	0.948 ± 0.003	0.947 ± 0.003
	out-of-distribution (non-contrast CT)	0.207 ± 0.209	0.860 ± 0.009	0.873 ± 0.015	0.887 ± 0.006
Spleen	in-distribution (contrast CT)	0.884 ± 0.029	0.890 ± 0.037	0.919 ± 0.005	0.904 ± 0.032
	out-of-distribution (non-contrast CT)	0.038 ± 0.009	0.654 ± 0.031	0.648 ± 0.051	0.691 ± 0.065
All Averaged	in-distribution (contrast CT)	0.916	0.924	0.935	0.932
	out-of-distribution (non-contrast CT)	0.101	0.535	0.529	0.747

*For definitions see section Experimental Setup. Mean ± sd.