Table 2.

Log-rank scores from a comparison of high/low standardized maximal distances (SDmax) for each method (Euclidian, Manhattan, Tchebychev, Travelling Salesman Problem distance, and Euclidian, Manhattan, Tchebychev voxels based) with the number of events, 4-year PFS (4y-PFS) and 4-year OS (4y-OS). Log-rank scores revealed significant differences in PFS and OS between high/low SDmax with all methods.

Method		SDmax _Euc	SDmax _Man	SDmax _Tch	SDTSP	SDmax _Euc_Vox	SDmax _Man_Vox	SDmax _Tch_Vox
Kappa	SDmax_Euc	NA	0.84	0.92	0.64	0.87	0.85	0.82
PFS	HR	2.90	2.70	3.11	2.95	3.58	2.84	3.66
	95% CI	1.8–4.7	1.7–4.2	2.0–4.9	1.7–5.1	2.2–5.8	1.8–4.5	2.2–6.1
	p-value	<0.0001	<0.0001	<0.0001	0.0001	<0.0001	<0.0001	<0.0001
	Se	0.27	0.32	0.30	0.18	0.25	0.30	0.23
	Sp	0.90	0.87	0.89	0.92	0.92	0.89	0.93
OS	HR	2.66	2.60	2.90	3.11	3.10	2.85	2.93
	95% CI	1.4–4.8	1.5–4.6	1.6–5.2	1.6–6.0	1.7–5.7	1.6–5.1	1.5–5.6
	p-value	0.0013	0.0009	0.0003	0.0008	0.0003	0.0004	0.0011
	Se	0.28	0.33	0.31	0.20	0.26	0.31	0.22
	Sp	0.88	0.84	0.87	0.91	0.89	0.87	0.91