Diffusion MRI (DMRI) is able to depict cerebral tissue microstructure in-vivo. Multi-Compartment (MC) models represent the DMRI signal as a weighted sum of components relying on pre-defined biophysical substrate and represented by parametric functions. Number and type of parameters depend on assumptions on the local properties of the tissue. Recent years have seen a proliferation of MC models1. The Spherical Mean Technique (SMT)2, exploiting spherical harmonics, factors out the neurite orientation distribution providing direct estimates of the structure. Moreover, the estimation of 5 parameters has shown not to be reliable because of the ill-posedness of the problem. In consequence, the value of the (same) microstructural descriptors are model- and instance-dependent. In addition, 5-shells acquisitions would be needed which is rare in real settings. In this work we characterize such effect on four simplified 2-parameters models.
Two-parameters compartmental models for diffusion MRI: a comparative analysis
M. Zucchelli
;L. Brusini
;G. Menegaz
2018-01-01
Abstract
Diffusion MRI (DMRI) is able to depict cerebral tissue microstructure in-vivo. Multi-Compartment (MC) models represent the DMRI signal as a weighted sum of components relying on pre-defined biophysical substrate and represented by parametric functions. Number and type of parameters depend on assumptions on the local properties of the tissue. Recent years have seen a proliferation of MC models1. The Spherical Mean Technique (SMT)2, exploiting spherical harmonics, factors out the neurite orientation distribution providing direct estimates of the structure. Moreover, the estimation of 5 parameters has shown not to be reliable because of the ill-posedness of the problem. In consequence, the value of the (same) microstructural descriptors are model- and instance-dependent. In addition, 5-shells acquisitions would be needed which is rare in real settings. In this work we characterize such effect on four simplified 2-parameters models.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
