Publication
Diffusion magnetic resonance imaging (dMRI) has demonstrated great capabilities to quantify tissue microstructure using biophysical diffusion models. Over the years, various gradient sequences have been proposed to provide sensitivity and specificity to various features of the tissue, using the pulse-gradient spin echo sequence (PGSI). In lymph node imaging, the long diffusion times required to estimate large characteristic length scales are achievable with the stimulated echo acquisition mode (STIAM). However, the butterfly gradients inherent to the STIAM protocol result in significant cross-terms in the diffusion-weighting matrix. If overlooked, these cross-terms can introduce biases in the microstructure estimations. Addressing these cross-terms within the analytical solution of the biophysical model is challenging. In this context, in-silico Monte-Carlo diffusion simulations in realistic numerical tissue substrates are a powerful tool to study diffusion without any assumptions on the analytical form of the signal. Using simulations and preclinical dMRI data, this work highlights (1) the STIAM bias over the PGSI sequence in free isotropic medium (agar), and (2) distinct regions of a rat lymph node correlate best with simulation in numerical substrates with distinct sphere densities. In summary, this work highlights the advantages of in-silico simulations for investigating MRI sequence effects on the DWI signal, yet it remains a challenging field.