Distinguishing Lymphomatous and Cancerous Lymph Nodes in 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography by Radiomics Analysis

Table 6

Two examples of how predictive models work (continued Figure 2).

Radiomic parameters (only 12 of the 100 parameters were shown, based on the predictive model—lymphoma versus cancer)
Sample 1	CTvalue_min_CT; −155.00; NGLDM_Coarseness_PET; 0.16	HISTO_Entropy _log10_CT1; 0.54; FDGuptake _TLG; (mL)_PET; 5.35	HISTO_Entropy _log2_CT; 4.98; GLRLM_RLNU _PET; 17.90	GLCM_Entropy _log10_CT; 1.90; SHAPE_Volume (mL) _PET; 1.21	GLCM_Entropy _log2_CT; 6.32; SHAPE_Volume (vx) _PET; 19.00	GLZLM_ZLNU_CT; 67.00; GLZLM_LZHGE _PET; 467.00
Sample 2	CTvalue_min_CT33; 0.00; NGLDM_Coarseness_PET; 0.06	HISTO_Entropy _log10_CT; 0.57; FDGuptake_TLG (mL)_PET; 36.80	HISTO_Entropy _log2_CT; 1.89; GLRLM_RLNU _PET; 86.30	GLCM_Entropy _log10_CT; 1.14; SHAPE_Volume (mL)_PET; 5.67	GLCM_Entropy _log2_CT; 3.80; SHAPE_Volume (vx)_PET; 89.00	GLZLM_ZLNU_CT; 29.50; GLZLM_LZHGE _PET; 798.00
Prediction calculated predictive variables according to the predictive model—lymphoma versus cancer
Sample 1			Sample 2
Predictive variable	Predictive scores (calculated from the predictive model)	Result	Predictive variable		Predictive scores (calculated from the predictive model)	Result
SUVmax	6.41	CLN	SUVmax		12.00	LLN
PREct	0.29	CLN	PREct		0.12	CLN
PREpet	0.18	CLN	PREpet		0.49	LLN
PREcombination	1.00	LLN	PREcombination		0.18	CLN
Patient’s disease type	LLN		Patient’s disease type			CLN

Abbreviations: CLN, cancerous-lymph nodes; LLN, lymphomatous-lymph nodes; HISTO, histogram; GLCM, gray-level co-occurrence matrix; GLRLM, gray-level run-length matrix; HGRE, high gray-level run emphasis; GLZLM, gray-level zone-length matrix; HGZE, high gray-level zone emphasis; RLNU, run-length nonuniformity; LZHGE, long-zone high gray-level emphasis; min, minimum; _CT, this parameter is extracted from the CT image; _PET, this parameter is extracted from the PET image; PREct, CT-predictive variables; PREpet, PET-predictive variables, PREcombination, the combination of PET and CT predictive variables.