Recently methods to measure axon diameter distributions using magnetic resonance imaging (MRI) diffusion spectroscopy have been developed. Most early methods used single diffusion encoding sequences such as pulsed gradient spin echo and are thus sensitive to axons of diameters larger than 5 μm. We previously simulated oscillating gradient (OG) spin echo sequences to study smaller axons which include the majority of axons constituting cortical connections. Here we test the use of OG to infer axon diameters on regions of interest within a human brain autopsy sample containing the corpus callosum, the ependymal layer, and the cortex. To our knowledge, this is the first study using OG temporal diffusion spectroscopy for inferences of human axon diameters.
A portion of the normal-appearing corpus callosum from an autopsy of human brain was kept in formalin before being prepared for imaging. 24 hours before MRI it was replaced with PBS. For imaging it was subsequently placed into a sample tube filled with agarose (2% w/v) in a 15 mL sample tube.
The sample was imaged using a 7 T MRI. Each 20 ms sinusoidal gradient pulse ranged from n = 1 to 15, in steps of 1. Six gradient strengths were used for each frequency and gradient pulses were separated by 24.52 ms.
Signals were fitted using two models, a two compartment ActiveAx model and AxCaliber model, using least squares minimization and mean axon diameters were extracted.
Ten 2000x magnification electron microscopy slices were studied from the region above the corresponding MRI. Axon diameters were manually measured using ImageJ software by drawing lines across the smallest diameter of cells that could be identified as axons.
Using the ActiveAx model, the mean effective diameter for axons in the corpus callosum ranged from: 1.8 ± 0.1 $\mu$m to 2.34 ± 0.05 $\mu$m. The mean effective diameter for ependymal cells in the ependymal layer ranged from 2.58 ± 0.06 $\mu$m to 2.80 ± 0.06 $\mu$m. Using the AxCaliber model, the axon diameter distribution indicated size of axons in the range of 0.9 to 3 $\mu$m. Our measurements made using histological techniques found 0.1-6 μm axon diameters in the sample, with an average overall of 0.6 ± 0.5$\mu$m, which is consistent with other histology measurements.
Our preliminary results indicate our method is sensitive to axons in the 2 $\mu$m range which is smaller than previous measurements. These axons constitute the majority of cortical connections making our measurements clinically relevant. Higher gradient frequencies will be needed for future studies to probe even smaller diameters and more work needs to be done to optimize the choices of gradient amplitudes, gradient frequencies, and SNR to make a method which can be used on live animals. This is the first step toward inferences of micron-sized axon diameters in the human corpus callosum.