Robustness of radiomics features of virtual unenhanced and virtual monoenergetic images in dual-energy CT among different imaging platforms and potential role of CT number variability

Insights into Imaging

Table 3 Inter-platform reproducibility of radiomics between each platform within the VUE images and within the VMI_70keV

Feature class	ICC > 0.90		ICC mean		CCC > 0.90		CCC mean
Feature class	VUE (%)	VMI_70keV (%)	VUE	VMI_70keV	VUE (%)	VMI_70keV (%)	VUE	VMI_70keV
First order (19 features)	45.03	49.36	0.6654	0.6811	44.68	49.36	0.6584	0.6749
Texture (75 features)	1.13	1.57	0.2324	0.2782	1.04	1.57	0.2247	0.2695
GLCM (24 features)	1.39	2.96	0.2871	0.3146	1.39	2.96	0.2782	0.3050
GLDM (14 features)	1.27	0.95	0.2093	0.2724	1.11	0.95	0.2028	0.2645
GLRLM (16 features)	0.97	0.97	0.2083	0.2644	0.69	0.97	0.2009	0.2560
GLSZM (16 features)	0.69	0.97	0.1963	0.2455	0.69	0.97	0.1895	0.2377
NGTDM (5 features)	1.33	0.44	0.2262	0.2681	1.33	0.44	0.2181	0.2588
Overall (94 features)	10.00	11.23	0.3199	0.3596	9.86	11.23	0.3124	0.3515

Percentage indicates the percentage of features met the cutoffs for robustness measures (ICC > 0.90 and CCC > 0.90). GLCM Gray-level co-occurrence matrix, GLDM Gray-level dependence matrix, GLRLM Gray-level run-length matrix, GLSZM Gray-level size zone matrix, NGTDM Neighborhood gray-tone difference matrix