18F‑FDG PET/CT based radiomics features improve prediction of prognosis: multiple machine learning algorithms and multimodality applications for multiple myeloma

Table 2 Finalized features screened from the step 3 incorporated in the model construction

Feature_names	VIF	HR	P-value
Adverse prognostic cytogenetics status	2.633	2.197	0.00018
RISS_Staging	3.645	2.482	0.00009
Red blood cells	4.372	0.613	0.01988
Albumin	4.173	0.447	0.00009
B2M	3.687	1.426	0.00472
Absolute Neutrophils	2.336	0.597	0.03329
wavelet.LLL_firstorder_10Percentile_PET	2.48	1.796	0.00396
exponential_glcm_InverseVariance_PET	2.766	1.995	0.00068
lbp.2D_firstorder_InterquartileRange_PET	4.352	1.470	0.00569
lbp.3D.k_glcm_MCC_PET	2.307	1.479	0.03082
original_shape_MajorAxisLength_PET	4.638	1.777	0.00408
original_shape_MinorAxisLength_CT	3.794	1.704	0.00233
original_ngtdm_Busyness_CT	3.335	1.378	0.00529
exponential_glszm_SmallAreaLowGrayLevelEmphasis_CT	1.52	0.492	0.00273
wavelet.HLH_gldm_SmallDependenceEmphasis_CT	1.765	0.552	0.01432

Riss Revised-International Staging System based staging, B2M β2-microglobulin, ISS International Staging System, HCT Hematocrit, A/G ratio Albumin/ Globulin (Ratio), GFR Glomerular filtration rate, LDH Lactate dehydrogenase
Adverse prognosis cytogenetics Status: FISH analysis confirmed del(17p), t (4;14), t (14;16), add 1q21

ISSN: 1471-2342