Improving high-resolution Q-Ball imaging with a head insert gradient: Bootstrap and SNR analysis

Introduction. The ability to depict small structures of the central nervous system using diffusion-weighted (DW) magnetic resonance imaging (MRI) is notably limited by the spatial resolution and sensitivity of the diffusion-encoded images. However, voxels smaller than ~6 mm (1.8mm isotropic) are difficult to achieve in conventional clinical scanners given the limited signal-to-noise ratio (SNR) of DW images and the amount of susceptibility artifacts when using single-shot acquisitions with a large matrix. A head insert gradient provides higher maximum gradient strength (2x than the regular system), higher slew rate, higher duty cycle and does not induce B0-drift. Using a head insert diffusion-encoding gradient areas can be achieved in less time and read-outs can be shortened thereby decreasing the echo time (TE) and yielding significant gains in SNR for comparable b-values. Additionally, the faster EPI readout reduces susceptibility distortion. Although head gradients have been used previously [1], they have not been combined with highly parallel detection (32ch receive coils) since many RF coils do not fit in the smaller insert gradient bore geometry. In this study we compare the performance of a head-insert gradient with a whole body gradient system using highly parallelized RF coils, for high resolution Q-Ball imaging. Bootstrapbased metrics demonstrate higher reproducibility of the Q-Ball orientation distribution functions (ODF) from head gradient HARDI data.