Anisotropic non-rigid Iterative Closest Point Algorithm for respiratory motion abdominal surface matching

Table 1 Quantitative evaluation of non-rigid isotropic and anisotropic ICP for real data

	Initial distance [mm]	Isotropic			Anisotropic			Anisotropic with landmark weighting
	Initial distance [mm]	M1 [mm]	M2 [unit]	M3 [%]	M1 [mm]	M2 [unit]	M3 [%]	M1 [mm]	M2 [unit]	M3 [%]
C1	17.4	3.4E−08	2.3	36.1	4.68E−08	2.78	95.88	4.59E−08	1.93	84.76
C2	8.0	1.3E−08	2.8	59.1	1.43E−08	3.48	96.44	1.49E−08	1.26	78.06
C3	7.0	4.1E−09	5.7	67.4	4.34E−09	5.74	95.83	3.79E−09	4.37	78.13
C4	11.7	2.4E−08	2.1	52.1	2.59E−08	1.18	93.28	2.18E−08	1.04	84.39
C5	8.0	1.7E−08	2.9	59.0	1.9E−08	2.89	94.5	1.92E−08	1.89	78.00
C6	7.6	1.5E−08	3.7	67.4	1.39E−08	3.11	96.44	1.29E−08	1.89	83.20
C7	5.9	1.2E−08	3.3	78.4	1.14E−08	3.30	95.91	1.14E−08	3.04	79.32
C8	5.8	1.4E−08	1.9	78.6	1.5E−08	2.48	98.86	1.42E−08	2.52	96.59
C9	6.7	7.4E−09	5.9	76.08	7.33E−09	6.22	96.17	8.3E−10	3.89	73.92
Median		1.4E−08	2.9	67.36	1.43E−08	3.11	95.91	1.42E−08	1.93	79.32

The case were M2 and M3 values for Anisotropic with landmark weighting are worse than the corresponding values for the other versions of the tested method have been marked as italics

ISSN: 1475-925X