Limitations of two-component diffusion models for axon diameter density estimation

Two component models of diffusion-weighed signal decay in Magnetic Resonance Imaging have been advocated for use in the estimation of axon diameter distributions. In this work, we demonstrate that axon diameters are not distinguishable under the short pulse approximation, even at high gradient strengths available on pre-clinical MRI systems. We instead investigate the long pulse regime under which axon diameters are maximally separated, as are the two hindered and restricted diffusion components. Through simulation and experimental MRI of the ovine optic nerve, we show that a simplistic two-component model is incapable of capturing experimental decay behaviour, calling into question the utility of these models for axon diameter density estimation.