SMNET: A Spherical Mean Network for Probing Tissue Microarchitecture from Single-Shell DMRI

Multi-shell diffusion MRI (ms-dMRI) allows for the examination of complex microstructure profiles and tissue changes. However, ms-dMRI is often limited in clinical applications due to its long acquisition time and dense q-space sampling requirements. In this study, we present a method that utilizes single-shell diffusion MRI (ss-dMRI) to characterize tissue microstructure. Our approach involves predicting multi-shell spherical mean images (ms-SMI) from ss-dMRI data and then employing advanced diffusion modeling to estimate the tissue microstructure. To this end, we combine a biological model with a data-driven deep learning approach to learn mappings between commonly co-acquired structural MRI and ss-dMRI to ms-dMRI. We validate the effectiveness of our method using an in vivo dataset in different health conditions collected by a clinical scanner. This work sets the foundation for future techniques that integrate clinical data acquisition protocols and enhance sensitivity to microstructure measures.