Table 3.
Results of multivariable logistic regression analysis in the three models.
| Models | Included features | Odds ratio (95% CI) | p value | Coefficient | Intercept |
|---|---|---|---|---|---|
| Model 1 | Radiomics signature | 1.004 | |||
| DISCRETIZED_HISTO_ExcessKurtosis | 3.160 (1.477–7.228) | 0.004 | 1.151 | ||
| GLRLM_GLNU | 0.180 (0.073–0.395) | <0.001 | -1.716 | ||
| GLRLM_RLNU | 284.479 (26.676–4552.497) | <0.001 | 5.651 | ||
| NGLDM_Coarseness | 5.813 (2.453–15.202) | <0.001 | 1.760 | ||
| Model 2 | Conventional CT images and clinical data | -1.989 | |||
| Size(mm) | 1.333 (1.160–1.533) | <0.001 | 0.288 | ||
| CTmean | 0.969 (0.949–0.990) | 0.003 | -0.31 | ||
| Model 3: | Model 1 + Model 2 | 0.6949 | |||
| PET_Rad_score | 3.073 (2.018–4.679) | <0.001 | 1.1227 | ||
| Size (mm) | 1.118 (0.955–1.309) | 0.1659 | 0.1114 | ||
| CTmean | 0.948 (0.926–0.971) | <0.001 | -0.0531 |
PET_Rad_score, PET radiomics signature; HISTO, histogram; GLRLM, gray-level run-length matrix; GLNU, gray-level non-uniformity for run; RLNU, run length non-uniformity, NGLDM, neighborhood gray-level different matrix; CI, confidence interval.
The bold values refer to the intercepts for calculation formula.