Improved Leakage Correction for Single-Echo Dynamic Susceptibility Contrast Perfusion MRI Estimates of Relative Cerebral Blood Volume in High-Grade Gliomas by Accounting for Bidirectional Contrast Agent Exchange.

BACKGROUND AND PURPOSE

Contrast agent extravasation through a disrupted blood-brain barrier potentiates inaccurate DSC MR imaging estimation of relative CBV. We explored whether incorporation of an interstitial washout rate in a leakage-correction model for single-echo, gradient-echo DSC MR imaging improves relative CBV estimates in high-grade gliomas.

MATERIALS AND METHODS

We modified the traditional model-based postprocessing leakage-correction algorithm, assuming unidirectional contrast agent extravasation (Boxerman-Weisskoff model) to account for bidirectional contrast agent exchange between intra- and extravascular spaces (bidirectional model). For both models, we compared the goodness of fit with the parent leakage-contaminated relaxation rate curves by using the Akaike Information Criterion and the difference between modeled interstitial relaxation rate curves and dynamic contrast-enhanced MR imaging by using Euclidean distance in 21 patients with glioblastoma multiforme.

RESULTS

The bidirectional model had improved Akaike Information Criterion versus the bidirectional model in >50% of enhancing tumor voxels in all 21 glioblastoma multiformes (77% ± 9%; P < .0001) and had reduced the Euclidean distance in >50% of enhancing tumor voxels for 17/21 glioblastoma multiformes (62% ± 17%; P = .0041). The bidirectional model and dynamic contrast-enhanced-derived kep demonstrated a strong correlation (r = 0.74 ± 0.13). On average, enhancing tumor relative CBV for the Boxerman-Weisskoff model exceeded that for the bidirectional model by 16.6% ± 14.0%.

CONCLUSIONS

Inclusion of the bidirectional exchange in leakage-correction models for single-echo DSC MR imaging improves the model fit to leakage-contaminated DSC MR imaging data and significantly improves the estimation of relative CBV in high-grade gliomas.