Meeting Banner
Abstract #3128

Using in-vivo Human Brain Data to Select Diffusion MRI Compartment Models

Uran Ferizi1, 2, Torben Schneider2, Eleftheria Panagiotaki1, Gemma Nedjati-Gilani1, Hui Zhang1, Claudia Angela M. Wheeler-Kingshott2, Daniel C. Alexander1

1Centre for Medical Image Computing and Department of Computer Science, University College London, London, England, United Kingdom; 2NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Institute of Neurology, London, England, United Kingdom

Finding models that are more descriptive than the standard Diffusion Tensor is necessary in producing better disease biomarkers that provide more specificity to the physiological changes and sensitivity to the pathological impact of the disease in the body. This work investigates which compartment models of diffusion MRI are best at describing the signal from in-vivo human brain white matter, and how reproducible these results are across acquisition sessions.This study helps clinicians and medical physicists in choosing models for future in-vivo brain microstructure imaging.

Abstract PDF Presentation Video

Keywords

acquisition analytical angular approval assessed assume audience averaging axis axon axons ball berg best better biological biophysical bovine brain brains clinical coherent come comes compartment compartments complex complexity computer computing conditions cook corpus create cylinder dataset datasets describing descriptive determine diameter diffusion dispersion distribution england enhance experiment explain field fitting fixed fluid full future generate goodness hall highest hindered human improve in vivo include isotropic isotropically jackknife kingdom known legend linear lowest mapping mathematical matrices matrix measuring minor model models much necessary neurology nine noise observations open oriented outer parker physicists pore previous principal processing protocol pulses queen radius random rank ranking reduced repeat resolution restricted restriction robust samples satisfying scanner science score segmenting select separate sessions shell shells significantly simpler simplicity smashing smith software source spectroscopy sphere springer stability stable stick sticks stop subjects subsets suppression target taxonomy tensor thick tissue uncertainty unit variations volume want white worst wright zeppelin zero zoom