Mutlimodality MR imaging for differentiation between brain tumor lesions

Purpose: Applying diffusion and perfusion metrics for evaluation of low-(LGG), high grade glioma (HGG) and metastases (MET) for differential diagnosis. Materials and Method: 43 patients (18HGG, 10 LGG, and 15MET) were included. MR data for tumour volume, perilesional edema, rCBF-, rCBV-, FLAIR-, FA-, ADC-maps were quantified by regions of interest (ROI). Measures of different parameters, and ratios, using contralateral white matter as denominator, were performed. A binary logistic regression model was constructed for multi-parametric analysis and ROCanalysis. Results: Significant difference was found for nADCt, rCBF, rCBV between LGG and HGG, nADCe between HGG and MET, and Ev, Ev-Tv ratio, nADCt, nADCe, rCBF, rCBV between LGG and MET. ROCanalysis for HGG compared to LGG showed 80 % sensitivity and 81.2 % specificity for nADCt, 100 % sensitivity and 100 % specificity for rCBF and 80 % sensitivity and 90 % specificity for rCBV. ROC-curves betweenMETand LGG showed sensitivity and specificity for Ev 73.3 % and 90 %, Ev-Tv ratio 80 % and 100 %, nADCt 90 % and 86.7 %, nADCe 80 % and 90 %, rCBF 93.3 % and 100 %, and rCBV 60 % and 100 %. Combining Ev, Ev-Tv ratio, nADCt, nADCe and rCBV between METand LGG gave 93.3%sensitivity and 100%specificity. Combining nADCt and rCBV between HGG and LGG 86.7 % sensitivity and 100 % specificity. Conclusion: Multi-parametric imaging protocols is an advantage for preoperative distinction of LGG, HGG and MET