Impact of the mono-exponential signal decay approximation on the numerically predicted spatial BOLD specificity for spin echo sequences

To study the BOLD effect numerically, blood vessels are often modelled as infinite cylinders [1] (infinite cylinder model, ICM). To simplify analytical calculations, signal dephasing resulting from spins diffusing around these infinite cylinders is then usually further approximated as a mono-exponential signal decay as suggested by Ogawa et al.[1]. Although for a wide range of parameters this mono-exponential approximation (MEA) is a good approximation for the signal resulting from the ICM it is known to increasingly deviate for increasing radii of the cylinders and when using spin echoes instead of gradient echoes [2]. In high resolution fMRI studies, however, spin echoes are sometimes used to increase the ratio R between the BOLD signal stemming from micro-vascular and macro-vascular tissue and thus increase the spatial specificity of the acquired activation (see e.g. [3]). For spin echoes, this ratio R has recently been investigated by Uludag et al. [2] for a wide parameter range using an analytical model for the BOLD signal which relied on the aforementioned MEA. To investigate the impact of the MEA on the predicted BOLD spatial specificity, we implemented the ICM into our general purpose MRI simulator JEMRIS [4]. JEMRIS solves the resulting Bloch equations using the numerical ordinary equation solver CVODE [5] and is thus capable of calculating the complete signal resulting from the ICM without further approximations such as the MEA.