LAUSR

Neuroimaging offers a non-invasive means to probe tumor tissue in order to inform decision making at all phases of brain tumor treatment. Diffusion MRI is particularly sensitive to tumor tissue microstructure, with greater heterogeneity being reflected as a larger diffusional kurtosis. Q-Space Trajectory Imaging (QTI) uses tensor-valued diffusion encoding (encoding along multiple directions per shot) to disentangle Isotropic Mean Kurtosis (MKi) from Anisotropic Mean Kurtosis (MKa), which are otherwise conflated in the Total Mean Kurtosis (MKt). To test whether disentangling MKi and MKa facilitates a more specific probe of tumor tissue heterogeneity and malignancy, we investigated if QTI parameters could distinguish low- from high-grade gliomas and enhancing from non-enhancing regions using pre-operative QTI imaging of 13 W.H.O. grade I-II and 18 grade III-IV glioma patients. We also analyzed these features separately for de novo and recurrent tumors. Regions of Interest (ROIs) were drawn on QTI maps, with support from T1 and T2-weighted images, for enhancing region, non-enhancing region, necrotic cavity, cyst, edema and resection cavity. ROC was used to gauge QTI parameter performance in classifying tumor characteristics. MKi was found to be the strongest predictor of tumor grade (AUC = 0.74, p = 0.019). MKi and MKt separated de novo from recurrent tumors (AUC = 0.80 and 0.76, p < 0.05). MKt separated enhancing regions from non-enhancing regions with an AUC of 0.88 in all tumors and 0.97 in de novo tumors. Our preliminary results highlight that tensor-valued diffusion MRI and QTI analysis have the potential to non-invasively characterize tumor grade. Further, MKt accurately differentiated enhancing from non-enhancing tumors and could potentially substitute for gadolinium injection in some situations thereby decreasing risk, time, and cost. Our ongoing studies in larger groups aim to further correlate molecular tumor markers (eg. IDH1 status) with these diffusion parameters.