A New Acquisition Protocol for Optimized Dynamic Susceptibility Perfusion Imaging of Brain Tumors

Take-Away Points

• ■ Major Focus: This study describes a technique for creating synthetic dynamic susceptibility perfusion MRI with multiple simulated acquisition parameters using a single contrast media bolus and acquisition.

• ■ Key Results: Prospective comparison of the synthetic technique in 38 patients with gliomas with external cohorts of glioblastoma and central nervous system (CNS) lymphoma demonstrated statistically significant differences in dynamic susceptibility perfusion MRI metrics among the tumor types.

• ■ Impact: Synthetic dynamic susceptibility contrast (DSC) perfusion imaging obtains different perfusion metrics using a single acquisition and contrast media bolus rather than relying on separate optimized acquisitions as done previously. Future opportunities include retrospective comparisons across heterogeneous cohorts by synthetically matching acquisition parameters.

DSC perfusion MRI is a recently developed functional imaging technique used to help distinguish different types of brain tumors, including gliomas, CNS lymphomas, and brain metastases. DSC perfusion MRI measures rapid changes in magnetic properties (susceptibility) in an area of interest following injection of a bolus of intravenous contrast media. DSC helps detect two parameters relevant for brain tumors: regional cerebral blood volume (CBV), which measures the relative vascularity of tissue, and percentage of signal recovery (PSR), a secondary metric for the underlying microvascular and cellular architecture of tissue. Optimal estimation of these two metrics requires two separate acquisitions (and contrast media administrations), as the two rely on different T1 and T2* weighting. PSR helps differentiate CNS tumor types on the basis of their underlying microarchitecture, while CBV may differentiate recurrent tumor from posttreatment changes.

Sanvito et al present a strategy to generate “synthetic” DSC signals that can be optimized for both CBV and PSR by using a dynamic spin- and gradient-echo echoplanar imaging (SAGE-EPI) acquisition and a single injection of contrast media. Their processing pipeline uses SAGE-EPI to disentangle and quantify the voxelwise T1 and T2* relaxation rates over the acquisition period. They then use nuclear magnetization Bloch equations to simulate the voxelwise signals that could be obtained with different acquisition parameters. The authors generated two separate synthetic CBV-optimized acquisitions (with and without a simulated preload bolus) and a PSR-optimized acquisition from a prospective cohort of 38 patients with histopathologically confirmed glioma who underwent SAGE-EPI scans. Estimation of PSR from their two synthetic CBV acquisitions correlated poorly with the synthetic PSR acquisition, matching prior work showing the importance of a PSR-optimized acquisition. Synthetic PSR values in patients with glioblastoma closely matched values from PSR-optimized acquisitions in an external cohort of patients with glioma, while they were different from another cohort of primary CNS lymphoma. This study shows the feasibility of the synthetic technique to accurately estimate PSR without requiring a dedicated PSR-optimized acquisition. Furthermore, because SAGE-EPI allows simulation of multiple different acquisition parameters, standardized comparison of perfusion metrics across heterogeneous multicenter cohorts could be possible.

Overall, this work presents an important technical advance in DSC acquisition. Several potential directions for future work include the effect of simulated preload adjustment on transverse relaxation rate and prospective SAGE-EPI acquisition of datasets from patients with primary CNS lymphoma.

Highlighted Article

• Sanvito F, Yao J, Cho NS, Raymond C, Telesca D, Pope WB, Everson RG, Salamon N, Boxerman JL, Cloughesy TF, Ellingson BM. “Synthetic” DSC perfusion MRI with adjustable acquisition parameters in brain tumors using dynamic spin-andgradient-echo echoplanar imaging. AJNR Am J Neuroradiol 2024; Sept 12. doi: https://doi.org/10.3174/ajnr.A8475