Biexponential modeling of multigradient-echo MRI data of the brain

Functional MRI (fMRI) using fast multigradient-echo acquisition methods allows the quantitative determination of the relevant parameter T2*. Previously, the TE-dependent signal decay has been modeled with a monoexponential function despite the complex composition of the brain. In this study, biexponential modeling was used to evaluate the relaxation of brain parenchyma and blood separate from that of cerebrospinal fluid. Single-shot multigradient-echo data acquired with spiral or EPI techniques were analyzed. In phantom experiments the biexponential method proved to be accurate. Compared to the biexponential procedure, the monoexponential model overestimated T2* (72.2 msec vs. 65.3 msec) and underestimated DeltaT2* (2.96 msec vs. 3.19 msec) during visual stimulation. The biexponential method may allow intrinsic correction for partial volume effects due to cerebrospinal fluid. The activation-induced parameter changes are detected with a sensitivity equal to that of a monoexponential method. The resulting T2* and DeltaT2* values describe the experimental data more accurately.