Computed diffusion contribution in the complex blood oxygenation-level dependent fMRI signal

Abstract

Diffusion is a ubiquitous phenomenon in blood oxygenation-level dependent (BOLD) functional magnetic resonance imaging (fMRI), but it cannot be completely separated out from BOLD signals experimentally or theoretically. Nevertheless, it is possible to numerically characterize the diffusion-only contribution by Monte Carlo simulation and signal analysis. The simulation was carried out for a variety of parameter settings of cortical vasculature (vessel radii =[2,16] micron) and field strength (B 0 = [0.18, 12.9] T). A cortical voxel was simulated in a cube of size 320 × 320 × 320 micron3 filled with a random vascular network. Diffusion-present and diffusion-absent signals were calculated by intravoxel dephasing (via switching on and off the diffusion coefficients) for a range of echo times T E =[0,60] ms. Given a pair of diffusion-present and diffusion-absent complex signals, the diffusion-only contribution was calculated by two signal analysis models: multiplication and addition. Statistics were rendered through multiple random realizations for each vasculature setting. Simulations suggest that diffusion contributes to BOLD signal (for setting of vessel radius = 3 micron, B 0 = 3 T, T E = 30 ms) by a signal amplitude decay rate change of 0.8 Hz (multiplication model) or by a signal amplitude change percentage of 5% (addition model). Overall, this contribution is maximal amid a large range of field strengths, diminishing toward low and high fields. © 2012 Wiley Periodicals, Inc.