Even if a bit late, I'm pleased to introduce the official twitter digest of our recent @Neuroimage_EiC paper on SANDI: Soma And Neurite Density Imaging ( https://doi.org/10.1016/j.neuroimage.2020.116835), joint work with @AndradaIanus @GuerrMic @luso_dnunes @fishpiechicken @ShemeshL @garyhuizhang(1/9)

We asked ourselves: do bodies of brain cells (from neurons to neuroglia) have an impact on the measured dMRI signal? And if so, can we measure it and disentangle it from other major sources like neurite and extra-cellular space? (2/9)

With numerical simulations, we showed that for simple PFG experiments soma has a specific signature at short diffusion times (<~20 ms) and high b (>3 ms/um2). Moreover, under these conditions, the exchange soma<->neurites can be neglected for most of the brain cell types (3/9).

These findings support the use of a simple non-exchanging three-compartment model (SANDI) to effectively characterize the main contributors to the dMRI signal in neural tissue: intra-soma (as sphere), intra-neurite (as sticks) and extra-cellular space (as Gaussian/tensor). (4/9)

This is a major departure from the current standard two-compartment model of neural tissue (5/9)

SANDI maps of apparent MR soma and neurite signal fraction showed a remarkable similarity with histological images of cell bodies and myelin staining, respectively. (6/9)

And we demonstrated the potential value of SANDI to characterise the brain cytoarchitecture with more specificity to soma density, in-vivo in 25 healthy human subjects from the @MGHMartinos Adult Diffusion Dataset. (7/9)

Finally, we found that, in ex-vivo mouse brain gray matter, the “sphere-compartment” introduced in SANDI provides a better fit of the dMRI signal at high b and more reasonable model parameters estimates than a “dot-compartment” counterpart. (8/9)

We are now working at further validation. Stay tuned for the upcoming histological validation. The preliminary results we presented at ISMRM 2019 ( https://discovery.ucl.ac.uk/id/eprint/10074393/1/Abstract_1_Palombo.pdf) were already pretty good (9/9)