Cardiac diffusion kurtosis imaging in the human heart in vivo using 300 mT/m gradients

Diffusion tensor imaging (DTI) is commonly used in cardiac diffusion magnetic resonance imaging (dMRI). However, the tissue's microstructure (cells, membranes, etc.) restricts the movement of the water molecules, making the spin displacements deviate from Gaussian behavior. This effect may be observed with diffusion kurtosis imaging (DKI) using sufficiently high b-values ( $b>450{\text{s/mm}}^2$ ), which are presently outside the realm of routine cardiac dMRI due to the limited gradient strength of clinical scanners. The Connectom scanner with $G_{\max }=300\mathrm{mT}/m$ enables high b-values at echo times (TE) similar to DTI on standard clinical scanners, therefore facilitating cardiac DKI in humans.

Cardiac-gated, second-order motion-compensated dMRI was performed with $b_{\max }=1350s/m{m}^2$ in 10 healthy volunteers on a 3T MRI scanner with $G_{\max }=300\mathrm{mT}/m$ . The signal was fitted to a cumulant expansion up to and including the kurtosis term, and diffusion metrics such as fractional anisotropy (FA), mean diffusivity (MD), mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK) were calculated.

We demonstrate deviation of the signal from monoexponential decay for b-values $>450s/m{m}^2$ ( $\mathrm{MK}=0.32\pm 0.03$ ). Radial kurtosis ( $\mathrm{RK}=0.35\pm 0.04$ ) was observed slightly larger than axial kurtosis ( $\mathrm{AK}=0.27\pm 0.02$ ), and the difference is statistically significant ( $\mathrm{RK}-\mathrm{AK}=0.08\pm 0.04$ , $p=2e-4$ ).

This work demonstrates the feasibility of quantifying kurtosis effect in the human heart in vivo (at an echo time shorter than typical TEs reported for cardiac DTI), using high-performance gradient systems (which are 4-8 times stronger than on standard clinical scanners). Our work lays the foundation for exploring new biomarkers in cardiac dMRI beyond DTI.