Differentiating Brain Metastasis and High-Grade Glioma Using Multi-b Value Diffusion MRI and Tumor Volumetry

To evaluate feasibility of multi-b value diffusion magnetic resonance imaging (MRI) and volumetry in differentiating between brain metastases and high-grade gliomas (HGGs) while producing a differentiation tool.

Preoperative brain MRI consisting of both morphological and multi-b value diffusion sequences of patients with HGGs and brain metastases was prospectively performed. Three-dimensional masks of enhancing and non-enhancing tumor and surrounding edema were semiautomatically segmented. Multiple diffusion parameters were subsequently derived together with volumes of the particular tissues. Histogram analysis of the diffusion parameters was performed, and the parameters' diagnostic power to differentiate between the subgroups was evaluated by receiver operating characteristic analysis and least absolute shrinkage and selection operator (LASSO) regression method.

A training dataset included 97 consecutive patients (67 HGGs, 30 metastases), whereas 17 patients (9 HGGs and 8 metastases) comprised a validation group. Overall, 66 histogram diffusion parameters and tissue volumes were found to differ significantly between metastasis and HGG subgroups. LASSO regression identified 17 of these as best predictors. A decision tree using four parameters achieved sensitivity of 90% and 87.5% and specificity of 97% and 77.8% for the training and validation subgroups, respectively.

Multi-b diffusion MRI and tumor volumetry may be valuable diagnostic tools for differentiating HGG from brain metastasis.