Differentiating high and low grade pediatric brain tumors using diffusional kurtosis imaging

Abstract

The purpose of this study is to determine the accuracy with which a non-Gaussian measure of diffusion, mean kurtosis (MK), predicts the histologic grade of pediatric brain tumors. After institutional review board approval, 21 World Health Organization (WHO) grade I, 7 WHO grade II, and 7 WHO grade IV pathologically-proven intracranial pediatric malignancies were retrospectively reviewed for preoperative diffusional kurtosis imaging. Multiple diffusion metrics of the tumors including MK, mean diffusivity (MD) and fractional anisotropy (FA) were determined. Comparisons between groups were performed using the Mann-Whitney test (p < .05). Receiver operating characteristics analysis was done to assess accuracy of each metric in predicting histologic grade. MK was significantly higher for grade IV neoplasms (0.97, p < 0.0004) than grade I (0.62) or grade II (0.67) tumors. MD was significantly higher for grade I (1.43) compared with grade IV neoplasms (1.07, p < 0.018), however not for grade II (1.43) compared with grade IV (p < 0.08) tumors. FA did not differ significantly between grades. Area under the receiver operating characteristic curve was highest for MK (0.94) and lower for MD (0.89). FA performed only slightly better than chance (0.54). MK is an accurate diffusion metric for predicting histologic grade of pediatric brain tumors, consistent with conclusions from prior studies demonstrating similar results in adult populations.

Keywords

Magnetic resonance imaging - diffusion - kurtosis - brain tumors