Table 2.

Differences of parameters between aGBM and PCNSL

Parameters	aGBM	PCNSL	p value	AUC (95% CI)	Cutoff	Accuracy	Sensitivity	Specificity
rADC	1.329(1.138, 1.471)	1.116 (1.023, 1.205)	< 0.001	0.751(0.641, 0.860)	1.325	0.811(107/132)	0.886	0.556
Ktrans	0.041(0.033, 0.050)	0.075 (0.052, 0.111)	< 0.001	0.808(0.734, 0.881)	0.065	0.674(89/132)	0.591	0.963
Kep	0.360(0.199, 0.510)	0.306 (0.229, 0.415)	0.37	0.556(0.426, 0.687)	0.404	0.674(89/132)	0.733	0.482
Ve	0.122(0.082, 0.147)	0.227 (0.148, 0.363)	< 0.001	0.765(0.657, 0.874)	0.141	0.780(103/132)	0.791	0.741
iAUC	0.079(0.062, 0.092)	0.129 (0.094, 0.184)	< 0.001	0.801(0.728, 0.875)	0.122	0.576(86/132)	0.562	1.000
rCBV	8.253(6.209, 11.113)	4.667 (3.809, 6.147)	< 0.001	0.817(0.708, 0.925)	6.171	0.765(101/132)	0.762	0.815
DCE-MRI*				0.827(0.752, 0.887)		0.735(97/132)	0.676	1.000
rADC + DCE-MRI*				0.877(0.808, 0.928)		0.758(100/132)	0.829	0.889
rADC + rCBV				0.936(0.880, 0.971)		0.879(116/132)	0.876	0.889
rCBV + DCE-MRI*				0.918(0.858, 0.959)		0.894(118/132)	0.924	0.815
rADC + DCE-MRI*+rCBV				0.966(0.919, 0.990)		0.932(123/132)	0.943	0.889

Note. — Data for glioblastoma and PCNSL are median values and numbers in parentheses are interquartile ranges; numbers in parentheses for AUCs are 95% confidence intervals (CI). P values were calculated with the t-test in rADC and rCBV, and the Mann-Whitney U test in Ktrans, Kep, Ve, and iAUC. The Ktrans and iAUC parameters have significant collinearity, and the DCE-MRI model excludes iAUC to prevent multicollinearity. DCE-MRI* means that the parameters Ktrans and Ve. aGBM = Atypical glioblastoma, DCE-MRI = Dynamic contrast-enhanced magnetic resonance imaging, iAUC = Initial area under the gadolinium concentration-time curve, Ktrans = Volume transfer constant, PCNSL = Primary central nervous system lymphoma, rADC = Relative apparent diffusion coefficient, rCBV = Relative cerebral blood volume, Ve = Volume of extravascular extracellular space per unit of tissue, Kep = Ktrans/Ve