Cerebrovascular reactivity (CVR) is increasingly recognized as a valuable clinical biomarker, making accurate, and reliable quantification essential, particularly in the absence of a gold-standard reference. However, both the acquisition and processing of CVR data are influenced by numerous methodological factors, including imaging technique, sequence parameters, vascular paradigm, and data processing strategies. These complexities can be daunting for newcomers and hinder methodological consistency across studies. To support both novice and experienced researchers, we systematically evaluated how different processing strategies affect CVR map repeatability, quantified using spatial intraclass correlation, across multiple vascular paradigms. Twenty-four healthy volunteers underwent BOLD–CVR imaging using CO 2 inhalation, breath-hold, and resting-state paradigms, each repeated in a test–retest setup. We found that optimal processing choices varied across paradigms and interacted in non-trivial ways. For example, the inclusion of motion-confounds and the application of temporal filtering require careful consideration, as they can introduce substantial collinearity in the regression model and reduce repeatability. We summarized these findings into practical insights to guide researchers in making sound methodological choices and promote consistency within the field.
Magnusson et al. (2026) studied this question.