Evaluation of prognostic models developed using standardised image features from different PET automated segmentation methods

EJNMMI Research

Table 4 Final output of prognostic models derived using AT, GCM3, KM2 and WT PET segmentation methods

	Parameter estimate	p value	Hazard ratio	95% CI
AT
Age	0.020	0.001	1.020	1.008–1.033
Treatment	− 1.075	< 0.001	0.341	0.254–0.459
Stage	0.144	< 0.001	1.155	1.072–1.245
GCM3
Age	0.019	0.003	1.019	1.006–1.032
Treatment	− 1.024	< 0.001	0.359	0.266–0.485
Stage	0.142	< 0.001	1.153	1.068–1.245
Kurtosis	0.632	0.002	1.882	1.260–2.809
Skewness	− 0.789	0.044	0.454	0.211–0.980
KM2
Age	0.020	0.001	1.020	1.008–1.033
Treatment	− 1.075	< 0.001	0.341	0.254–0.459
Stage	0.144	< 0.001	1.155	1.072–1.245
WT
Age	0.018	0.004	1.018	1.006–1.031
Treatment	− 1.063	< 0.001	0.345	0.257–0.464
Stage	0.140	< 0.001	1.150	1.065–1.242
GLNU	0.017	0.006	1.017	1.005–1.029
Skewness	0.674	0.030	1.962	1.067–3.607