Table 3. Per patient relative volume error for routine and corrected automatic segmentations.

	Routine			Corrected
	Rad	Res1	Res2	Res3	RA
Patient1†	8.89%	6.56%	11.73%	4.20%	0.70%
Patient2	8.52%	4.44%	1.39%	4.59%	3.87%
Patient3	1.84%	4.42%	2.45%	5.57%	5.41%
Patient4	1.15%	6.19%	2.07%	2.25%	4.56%
Patient5	4.72%	0.24%	3.47%	0.93%	0.93%
Patient6	2.81%	1.09%	4.47%	6.30%	6.62%
Patient7	3.49%	0.26%	1.17%	3.56%	3.32%
Patient8	3.51%	1.10%	3.37%	4.88%	3.91%
Patient9	2.27%	0.87%	0.09%	2.11%	2.62%
Patient10	22.62%	22.89%	22.18%	1.62%	0.09%
Patient11	7.24%	2.69%	2.12%	0.78%	0.87%
Patient12	3.10%	0.17%	2.40%	7.73%	7.73%
Patient13	2.24%	0.79%	5.97%	7.86%	6.72%
Patient14	6.06%	4.93%	4.61%	3.56%	2.54%
Patient15	11.39%	5.24%	6.50%	0.52%	0.43%
Patient16	5.07%	3.37%	4.29%	5.15%	4.51%
Patient17	1.13%	1.34%	4.21%	2.66%	2.62%
Patient18*	6.79%	1.17%	10.58%	2.71%	1.83%
Patient19	9.14%	4.28%	2.49%	1.57%	1.38%
Patient20	6.12%	1.42%	2.87%	1.51%	1.42%
Patient21	4.31%	2.30%	0.49%	3.91%	3.93%

Case with the highest inter-observer variability among manual routine* and corrected^† segmentations: 6.27% and 3.50%, respectively.