BLUE RIBBON POSTER IMAGING: Characterizing the Voxel-Wise Contrast to Noise Ratio of the Consensus Single-Dose Low Flip Angle DSC-MRI Perfusion Imaging Protocol
Abstract
Purpose
Single-dose, low flip angle (LFA) dynamic susceptibility contrast MRI (DSC-MRI) has been adopted as a consensus protocol for brain tumor perfusion imaging. However, a lower flip angle could reduce contrast-to-noise ratio (CNR) which may compromise voxel-wise precision and clinical interpretability. This study evaluates whether LFA DSC-MRI preserves diagnostic-quality voxel-wise CNR for cerebral blood volume (CBV) mapping in brain tumors across scanners and field strengths, while enabling reduced gadolinium dose.
Methods
DSC-MRI data are being curated as part of a larger multi-institutional database spanning two clinical sites, multiple scanner vendors, and different field strengths. The results presented here reflect a preliminary analysis of a single-site subset consisting of five patients with histologically confirmed glioblastoma who underwent sequential single-dose LFA and standard double-dose moderate-flip-angle (MFA) DSC-MRI acquisitions on a 3T clinical scanner. Tumor regions of interest were delineated on contrast-enhanced T1-weighted images and transferred to coregistered DSC-MRI data. Voxel-wise ΔR2* time series were computed with leakage correction, and CNR was calculated as the peak ΔR2* divided by the baseline standard deviation. Voxel-wise CNR distributions were compared between LFA and MFA, and the proportion of voxels exceeding the established reliability threshold (CNR ≥ 4) was quantified.
Results
Across all tumor voxels from five patients, LFA and MFA DSC-MRI at 3T achieved reliable voxel-wise CNR for CBV mapping (CNR ≥ 4) in 83% and 86% of voxels, respectively, with no statistically significant difference between protocols (p = 0.4442). Although absolute CNR values were lower with LFA, the fraction of voxels meeting the reliability threshold was preserved, demonstrating comparable voxel-wise CNR performance.
Conclusion
These preliminary results demonstrate that single-dose LFA DSC-MRI preserves diagnostic-quality voxel-wise CNR for CBV mapping in glioblastoma. Ongoing analysis of the full multi-institutional dataset will further assess the robustness and generalizability of these findings across scanners and clinical settings.