A simple method to scale the macromolecular pool size ratio for computing the g-ratio in vivo
Mara Cercignani1,2, Giovanni Giulietti2, Nick Dowell1, Barbara Spano2, Neil Harrison1, and Marco Bozzali2

1Clinical Imaging Sciences Centre, Brighton and Sussex Medical School, Brighton, United Kingdom, 2Neuroimaging Laboratory, Santa Lucia Foundation, Rome, Italy

Synopsis

previous work suggested that the myelin volume fraction (MVF) is proportional to the macromolecular pool size ratio, F, derived from Magnetization Transfer (MT). However the proportionality constant seems to be dependent on the specific MT model, and requires histological validation. Here we propose a simple method based to derive such a constant.

PURPOSE

The g-ratio, ie the ratio of the inner to the outer diameter of a myelinated axon, is an index of axonal myelination, related to the physiology and function of an axon. Recently, Stikov et al [1] proposed a method based on magnetization transfer (MT) and diffusion MRI (dMRI) to measure it non-invasively. The current model assumes that the myelin volume fraction (MVF) can be derived from to the size of macromolecular proton pool size ratio (F), estimated by quantitative MT, as MVF=αF, where α is a scaling factor. It is known, however, that α can vary depending on the specific MT method or model used [1] and requires histological validation. As this is not always an option, here we propose a simple approach to estimate α based on simulations.

METHODS

Data were collected at 2 sites, using 2 different scanners and 2 different imaging protocols. At site A, data from 23 healthy participants (M/F=10/13; mean age=38.6, SD=13.6yrs) were acquired on a 3T MRI scanner, including multi-shell diffusion-weighted data (10 b=0 volumes, 30 diffusion directions with b=710 smm-2, and 60 diffusion directions with b=2800 smm-2), optimised for neurite orientation dispersion and density imaging (NODDI, [2]), a series of 10 MT-weighted 3D gradient echo volumes with differing MT-weightings, a T1-mapping sequence, and a B1-mapping sequence. At site B, data from 17 healthy participants (M/F=7/10; mean age=25.7, SD= 6.7yrs) were collected at 1.5T, with a NODDI acquisition (10 b=0 volumes, 9 directions with b=300 smm-2 30 directions with b=800 smm-2, and 60 diffusion directions with b=2400 smm-2) and a quantitative MT protocol based on balanced steady-state free precession (bSSFP), as described in [3]. A T1-mapping sequence was also acquired. Quantitative MT data for both sites were analysed using in-house software, yielding a voxel-wise estimation of F. dMRI data were analysed using either the AMICO [4] (site A) or the NODDI [2] (site B) toolboxes, to yield maps of the volume of the intra-cellular water compartment (Vic) and of the isotropic component (Viso). Quantitative MT and NODDI data were non-linearly co-registered using ANTs 1.9.x to bring all maps into the same MNI space. The JHU white-matter tractography atlas, available with FSL, was used to extract unbiased masks of the forceps major and forceps minor (thresholded at 20%). The mean F, Vic and Viso of these 2 tracts were estimated for every participant and then used to estimate the g-ratio for scaling factors α ranging from 1 to 5, and averaged within site. The resulting g-ratio values were plotted against α to identify the value corresponding to the g-ratio ≈ 0.7 for either site [1]. The optimal α was then used to compute voxel-wise g-ratio maps and extract the values from the 5 sections of the central slice of the corpus callosum proposed by Hofer and Frahm [5]. The results were compared with estimates from histology and MRI in [1].

RESULTS

Fig 1 shows the simulated g-ratio values for the forceps major. A very similar plot was obtained for the forceps minor. The data suggest that a value of α=2.4 for site A and α=2.5 for site B provide a g-ratio of ~0.7. Fig 2 shows the g-ratio map of the corpus callosum from a different representative subject for either site. Fig 3A shows the mean g-ratios for the 5 callosum segments, averaged across subjects (per site). The values obtained in the cynomolgus macaque corpus callosum by Stikov et al [1] from MRI (top) and electron microscopy histology are shown for comparison (Fig 3B).

DISCUSSION

Our method provides a simple way of estimating α when histological validation is not available. Interestingly, we found similar α values for both sites, consistent with the value of 2.5 originally suggested by Dula et al [6]. Fig 3A indicate that the group values differ between sites, especially along the body of the callosum. This could be explained by the fact that the data were collected at different field strengths, and using very different MT acquisitions and modelling. The effects of noise and field uniformity are expected to affect g-ratio measurement. In addition, participants at site B were significantly younger than those at site A. Although the ranges of values are very similar, neither pattern in Fig 3A exactly matches that published in [1] and reported in Fig 3B; however, our data were obtained in humans (as opposed to macaque), and are group averaged. In addition, our values were averaged within 5 segments of the callosum instead of 8 portions.

Acknowledgements

No acknowledgement found.

References

1) Stikov et al. In vivo histology of the myelin g-ratio with magnetic resonance imaging. Neuroimage 2015, 4:368-73; 2) Zhang H, et al. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage. 2012, 61:1000-16. 3) Gloor et al., Quantitative magnetization transfer imaging using balanced SSFP. Magn Reson Med. 2008, 60:691-700; 4) Daducci et al., Accelerated Microstructure Imaging via Convex Optimization (AMICO) from diffusion MRI data. Neuroimage. 2015 Jan 15;105:32-44. 5) Hofer & Frahm, Topography of the human corpus callosum revisited--comprehensive fiber tractography using diffusion tensor magnetic resonance imaging. Neuroimage. 2006, 32:989-94.. 6) Dula AN, et al. Multiexponential T2, magnetization transfer, and quantitative histology in white matter tracts of rat spinal cord. Magn Reson Med. 2010 63:902-9.

Figures

Fig 1. G-ratio estimated from the mean F, Vic and Viso of the forceps major. Values are averaged across subjects, per site, and plotted against α. The bars indicate the cross-subject standard deviation

Fig 2. Single-subject g-ratio maps of the callosum from site A (top) and site B (bottom). The maps are in MNI space, from 2 different participants. Some common features can be spotted along the corpus callosum.

Fig 3. Mean g-ratio along 5 callosum segments [5] (A). For comparison the estimates obtained by Stikov et al [1] along 8 callosum segments of the cynomolgus macaque are also shown (B). EM=electron microscopy



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3369