A Multiatlas Segmentation Using Graph Cuts with Applications to Liver Segmentation in CT Scans

Table 1

Average values of the measures and scores for all 20 training images and 10 test images: volumetric overlap error (), relative absolute volume difference (), average symmetric surface distance (), root mean square symmetric surface distance (), and maximum symmetric surface distance ().

Type		[%]	[%]	[mm]	[mm]	[mm]
Initial; Solution	Measures	16.5 ± 3.8	8.6 ± 5.8	3.6 ± 1.6	7.1 ± 3.9	50.2 ± 24.8
	Scores	37	56	21	19	39	34.4
Affine prob.; Atlas	Measures	11.2 ± 2.8	2.3 ± 1.8	2.2 ± 0.9	4.4 ± 1.9	35.8 ± 15.5
	Scores	56	88	44	40	53	56.3
The proposed approach; 20 training images	Measures	5.2 ± 1.2	0.9 ± 1.2	1.0 ± 0.4	2.2 ± 1.1	26.9 ± 10.7
	Scores	79	91	76	70	65	76.3
The proposed approach; 10 training images	Measures	7.6 ± 3.2	−0.5 ± 3.9	1.3 ± 0.7	2.9 ± 1.8	24.7 ± 10.7
	Scores	70	87	68	60	68	70.5