Data-driven analysis of cerebrovascular reactivity mapping with breath-hold challenges in brain tumor patients

Motivation: Cerebrovascular reactivity (CVR) mapping can provide valuable information for the evaluation of lesion and neurovascular uncoupling in brain tumors. Goal(s): To evaluate the robustness of data-driven CVR analysis in patients with brain tumors. Approach: CVR MRI of brain tumor patients (n=18) was performed with a breath-holding task paradigm. CVR map was obtained using GLM with four regressors for comparison: (1) gray matter (GM), (2) GM with lesion removed, (3) whole brain (WB), and (4) WB with lesion removed. Results: Proper temporal and spatial filtering reduced the differences between the four regressors, resulting in similar CVR maps with highly correlated BOLD contrast. Impact: The study demonstrated that with proper temporal processing, robust data-driven CVR analysis can be obtained in patients with brain tumors.