Purpose: To present and quantify the performance of a method to compute tissue hemodynamic parameters from dynamic susceptibility contrast (DSC) MRI data in brain tissue with possible nonintact blood-brain barrier.
Theory and materials and methods: We propose a Bayesian scheme to obtain perfusion metrics, including capillary transit-time heterogeneity (CTH), from DSC-MRI data in the presence of contrast agent extravasation. Initial performance assessment is performed through simulations. Next, we assessed possible over- or under correction for tracer extravasation in two patients receiving contrast agent preloading and two patients not receiving preloading. Perfusion metrics for N = 60 patients diagnosed with either grade III (N = 14) or grade IV gliomas (N = 46) were analyzed across tissue types to evaluate the ability to distinguish regions with different hemodynamic patterns. Finally, N = 4 patient cases undergoing anti-angiogenic treatment are evaluated qualitatively for treatment effects. All patient data were acquired at 3.0 Tesla.
Results: The simulation studies showed good robustness against low signal-to-noise ratios, exemplified with Pearson correlations of R = 0.833 (mean transit time) and R = 0.738 (CTH) at signal-to-noise ratio = 20. Region-of-interest analysis of the N = 60 glioma patients showed that cerebral blood volume (CBV) significantly separated enhancing core from edema (grade IV: P < 10-8 , grade III: P < 0.05) and enhancing core from normal appearing ipsilateral white matter (NAWM) (grade IV: P < 10-8 , grade III: P < 0.05). The microvascular parameters were particularly good in separating edematous tissue from NAWM tissue in grade IV gliomas (P < 0.001). Finally, CTH separated grade III and grade IV core tissue (P < 0.05).
Conclusion: We have demonstrated robustness of the proposed Bayesian algorithm against experimental noise and demonstrated complementary value in microvascular parameters to the CBV parameter in separating tissue types in gliomas.
Level of evidence: 3 Technical Efficacy: Stage 2 J. MAGN. RESON. IMAGING 2017;46:537-549.
Keywords: Bayesian expectation-maximization; DSC-MRI; capillary transit-time heterogeneity (CTH); cerebral neoplasms; contrast agent leakage; oxygen extraction fraction (OEF).
© 2016 International Society for Magnetic Resonance in Medicine.