Combined gradient- and spin-echo EPI acquisition technique for high-resolution fMRI

INTRODUCTION – Gradient-echo (GRE) based fMRI pulse sequences are most sensitive to larger vessels, including veins. The regions-of-interest for neuronal activity, however, are mainly within capillaries. Effective localization of brain activity would require that the BOLD signal is confined to the microvasculature. However, in GRE-based sequences BOLD signal changes tend to be located distant to the sites of increased brain activity [1], causing misregistration between BOLD signal localization and the sites of underlying neuronal activity. In contrast, the spin-echo (SE) based pulse sequence [2] is a promising alternative which gives higher sensitivity to the microvasculature, but lower overall BOLD-sensitivity. In order to increase resolution in BOLD-fMRI, increasing spatial resolution of the acquisition does help, but the addition of SE-based image acquisition will enhance the resolution even further. Incorporating GREand SE-readouts into one pulse sequence combines the advantages of higher overall BOLD-sensitivity of the GRE-signal with better specificity of the SE-signal. The extension of a multi-echo gradient-echo pulse sequence to acquire both GREand SE-signal has been shown useful for vessel size imaging [3] and DSC-PWI [4]. A similar approach [5] was used for lowresolution BOLD-fMRI to estimate vessel size by combining the acquisition of GREand SE-signals. In this study, we apply a combined GRE-/SE-EPI pulse sequence to acquire high-resolution BOLD-fMRI images. By increasing spatial resolution in BOLD-fMRI, image acquisition will increase as well – leading to prolonged readout times. Without parallel imaging (PI) acceleration, a two-fold increase in spatial resolution will increase acquisition time by a factor of approximately four. Assuming mono-exponential signal decay, GRE-based fMRI is most sensitive to BOLD signal changes at TE ≈ T2*, and SE-based techniques are most sensitive at TE ≈ T2 [3,6]. At 3T, the T2* of brain tissue is around 35-40 ms, and T2 equals 90-100 ms, adding timing-constraints to high-resolution acquisitions. In the present study, we used PIacceleration to achieve shorter readout times, thus allowing the inclusion of a GRE-EPI readout between the 90° excitation pulse and the 180° SE-refocusing pulse.