LAUSR

Objectives To determine whether water kurtosis and diffusional metrics derived from diffusional kurtosis imaging (DKI) within primary central nervous system lymphomas (PCNSLs) and high-grade gliomas (HGGs) correlate with cellularity and/or nuclear-to-cytoplasmic (N/C) ratio. Methods Forty-four and 43 pathologically confirmed high-grade glioma and primary central nervous system lymphoma specimens between May 2013 and November 2016 were retrospectively reviewed. Diffusional metrics, kurtosis metrics, cellularity, and N/C ratios in PCNSLs and HGGs were compared using the Mann–Whitney U test (significant level, p < 0.007 [0.05/7]); Bonferroni correction). Results Mean kurtosis (MK), axial kurtosis (K // ), and radial kurtosis (K ⊥ ) in PCNSLs were 0.857 (0.693–0.924), 0.837 (0.660–0.941), and 0.834 (0.685–0.937), respectively; and 0.629 (0.524–0.716), 0.575 (0.511–0.689), and 0.675 (0.532–0.766), respectively, in HGGs (all p < 0.001). No significant differences in fractional anisotropy (FA), mean diffusion (MD), axial diffusion (λ // ), and radial diffusion (λ ⊥ ) were found between HGGs and PCNSLs. Cellularity was higher in PCNSLs than in HGGs ( p = 0.125); whereas, the N/C ratio in PCNSLs was significantly higher than that in HGGs ( p < 0.001). All DKI metrics (FA, MD, λ // , λ ⊥ , MK, K // , and K ⊥ ) were significantly correlated with N/C ratio in PCNSLs with correlation coefficients being rho = 0.418, − 0.722, − 0.525, − 0.768, 0.704, 0.579, and 0.686, respectively. Cellularity in PCNSLs and HGGs did not correlate with any kurtosis or diffusional metrics. Conclusions Difference of kurtosis parameters between PCNSLs and HGGs is correlated with their diverse N/C ratio. Key Points • DKI has considerable value in differentiating between PCNSLs and HGGs. • DKI can provide important information on nuclear-to-cytoplasmic ratio. • Difference of kurtosis parameters between PCNSLs and HGGs correlated well with their diverse N/C ratios.