Model-free leakage index estimation of the blood-brain barrier using dual dynamic susceptibility contrast MRI...

Pharmacokinetic K₂ mapping from dynamic susceptibility contrast (DSC)-MRI can be a sensitive technique for evaluating the vascular permeability of the subtly damaged blood-brain barrier (BBB) in ischemic regions. However, the K₂ values of ischemic lesions depend upon the selection of the intact BBB reference region. As previous observations suggest that the ΔR₂^* curve of pre-loaded DSC-MRI is not significantly affected by the extravasation of contrast agent, dual DSC-MRI acquisitions can be performed to derive the BBB leakage index from the voxel-wise reference input function for ischemic regions. This study aims to demonstrate the robustness of such model-free leakage index estimation in ischemic brains. By configuring the relationship between dual ΔR₂^* curves of the intact contralateral brain, the deviation of the measured ΔR₂^* curve from the unloaded DSC-MRI with respect to the non-deviated ΔR₂^* curve in the pre-loaded DSC-MRI can be quantified as the BBB leakage index. Such model-free leakage index values from rats with transient middle carotid artery occlusion (tMCAO) (n = 17) and normal controls (n = 3) were evaluated and compared with conventional K₂ values with multiple reference regions. Inter-subject leakage index values were also compared with the corresponding ΔT₁ map. Evans-blue-stained images were used to validate the leakage index. For the tMCAO group, leakage index values correlated well with ΔT₁ (Pearson's r = 0.828). The hyperintense area on the leakage index map matched well with the corresponding Evans-blue-stained area (Dice correlation = 0.626). The slopes of the scatter-plot from the leakage index (0.97-1.00) were observed to be more robust against changes in the reference region than those from conventional K₂ values (0.94-1.07). In a subtly damaged BBB tMCAO model, model-free evaluation of vascular permeability using dual DSC-MRIs would provide a consistent measure of inter-subject vascular permeability.