DYNAMIC MODELING OF CEREBRAL HEMODYNAMICS IN THE MRI SETTING

Abstract

The widespread use of magnetic resonance imaging (MRI) may offer an opportunity to evaluate cerebral hemodynamics at larger and more convenient scales. MRI also provides for the assessment of regional differences in hemodynamics within different brain regions. Our objective was to derive measures of cerebral hemodynamics (autoregulation and chemoregulation) using simultaneous recording of cardiorespiratory signals during resting state fMRI. It is hypothesis that the utilized modeling approach will be able to derive hemodynamic patterns similar to these previously described using traditional methods. Two models were designed: a single-output model defined by the global BOLD signal averaged over the entire brain, and a four-output model defined by the BOLD signal averaged over the frontal, parietal, occipital, and temporal lobes. Both models were defined by the three inputs, mean arterial blood pressure (MABP), pulse rate (PR), and end-tidal CO2 (ETCO2). Data were collected on 87 subjects (mean age 64, 56 female, 46 white, 39 black, 49 MCI). The models were considerably variable between the global and four regional BOLD signals. However, the global model for the PR and ETCO2 inputs were observed to exhibit a pattern similar to that previously reported. The model for the MABP input, unaccounted for in previous studies, exhibited a similar pattern to the putative autoregulatory response observed using TCD data. It was found that the indices derived to quantify PR reactivity was reduced in this dataset among those with MCI compared to those without MCI (global mean: normal = 0.03, MCI = −.02, p = 0.009; frontal mean: normal = 0.02, MCI = −.03, p = 0.007; parietal mean: normal = 0.02, MCI = −.01, p = 0.041; temporal mean: normal = 0.01, MCI = −.02, p = 0.053). The obtained three-input models exhibit characteristics similar to previous studies under the single-input constraint. The results suggest the potential utility of the dynamic modeling cerebral hemodynamics during rest in the MRI setting. The PR reactivity index may have clinical value in delineating individuals with increased risk for AD.