Multiexponential T2, magnetization transfer, and quantitative histology in white matter tracts of rat spinal cord

Abstract

Quantitative MRI measures of multiexponential T(2) relaxation and magnetization transfer were acquired from six samples of excised and fixed rat spinal cord and compared with quantitative histology. MRI and histology data were analyzed from six white matter tracts, each of which possessed unique microanatomic characteristics (axon diameter and myelin thickness, in particular) but a relatively constant volume fraction of myelin. The results indicated that multiexponential T(2) relaxation characteristics varied substantially with variation of microanatomy, while the magnetization transfer characteristics remained close to constant. The most-often-cited multiexponential T(2) relaxation metric, myelin water fraction, varied by almost a factor of 2 between two regions with myelin volume fractions that differed by only approximately 12%. Based on the quantitative histology, the proposed explanation for this variation was intercompartmental water exchange, which caused the underestimation of myelin water fraction and T(2) values and is, presumably, a greater factor in white matter regions where axons are small and myelin is thin. In contrast to the multiexponential T(2) relaxation observations, magnetization transfer metrics were relatively constant across white matter tracts and concluded to be relatively insensitive to intercompartmental water exchange.

Fig. 1

Schematic of rat spinal cord, identifying grey matter and six white matter tracts: Vestibulospinal (VST), Funiculus Cuneatus (FC), Rubrospinal (RST), Reticulospinal (ReST), Funiculus Gracilis (FG), and dorsal corticospinal (dCST).

Fig. 2

top) Light microscopic images (75 μm × 55 μm) from (left-to-right) dCST, FG, ReST, RST, FC, and VST. bottom) Example quantitative histology from FG and VST tracts: filtered and thresholded images for myelin fraction (MF) calculation, and manually defined myelin thickness (MyTh) and axon diameter (AxD) measurements.

Fig. 3

a) Typical T₂-weighted image showing manually drawn ROIs in grey matter and each of the six white matter tracts. b) MWF map. c) PSR map. The scale-bar on the right applies to both MWF and PSR.

Fig. 4

Typical T₂ spectra from 3 WM tracts (VST, FG, and dCST). Fine traces are spectra from 10 randomly chosen voxels from a given tract and the bold traces are the average of T₂ spectra from every voxel in a given WM tract.

Fig. 5

From top to bottom are plotted measures of: a) MWF/MF, b) PSR/MF, c) MWT₂, d) OWT₂, and e) (MWF/MF)×(OF/MWF) vs both axon diameter and myelin thickness. In each frame, triangle markers indicate the plots vs axon diameter and the top axis defines the appropriate scale. Similarly, diamond markers indicate plots vs and myelin thickness and the bottom axis defines the scale. Data from each tract are plotted with the color corresponding to their identification in Fig 1. In all cases, error bars are standard deviations across samples.

Fig. 6

Scatter plots of PSR vs MWF for all WM tract voxels from all six spinal cord samples, presented by WM tract. In each frame is shown the best fit linear function and correlation coefficient. In all cases, linear correlations were highly significant (p << 0.01) owing to the large number of data point (≈ 1000).

Fig. 7

top) A model two-pool system with volumes = 0.55 and 0.45, water proton densities of 0.44 and 0.82, intrinsic T₂s of 15 ms and 65 ms, and exchange defined by mean lifetimes in the short-lived pool (representing myelin) ranging from 40 ms to 500 ms. bottom). The T₂ spectra for five different mean myelin water lifetimes and the MWF values extracted from three of these spectra (indicated by matching color with legend).