To evaluate the utility of using texture heterogeneity in T2-weighted MRI to assess myelin integrity as compared to magnetization transfer ratio (MTR) and myelin water fraction (MWF), with validations by the corpus callosum in healthy mouse brain.

We acquired brain MR images from a 9.4T scanner (Brucker BioSpin, Ettlingen, Germany) from 7 healthy mice in vivo. Whole brain imaging was conducted for T2-weighted MRI using a RARE sequence (TR/TE = 4000/15ms; slice thickness = 0.5mm, 12 averages; matrix = 256x256, 25 slices) and for MT imaging (TR/TE =5000/15ms; MT power = 5 mT/m; pulse number = 60; matrix = 128x128; and slice thickness = 1mm). Single-slice multi-echo T2 MRI was acquired from the central location of T2 imaging (TR/TE = 3000/6ms; matrix = 128x128; slice thickness = 1mm, 128 echoes). Our measurements were focused on the corpus callosum (CC) as this is the largest inter-hemispheric white matter structure with highly organized white matter tracts. MTR was measured as the signal intensity ratio between images with and without off-resonance pulses [1], MWF as the fraction of water content with relaxation equal to or shorter than 40 ms using an open-source software [2] and texture heterogeneity as the similarity of multi-scale MRI spectra localized at individual image voxels using a spatial frequency-based approach [3]. Lower heterogeneity reflects greater coherency.

Imaging outcomes were recorded as per anatomy (genu, body and splenium) and hemispheric locations (right, center, and left) of the corpus callosum. Based on an established atlas [4] (Figure 1), we identified 5 slices in T2-weighted MRI to examine the corpus callosum: 1 slice for the genu, 3 for the body, and 1 for the splenium (Figure 2). Regions of interest (ROIs) were drawn along the boundary of the corpus callosum using ImageJ (NIH, USA) in selected image slices. Identified image and ROIs were then matched to the corresponding locations in MTR and MWF images for quantification. Outcome differences were evaluated using a mixed-effect modeling method that accommodated the variances both between animals and between imaging slices within an animal. For MWF, only differences between hemispheric locations were evaluated as only one slice was acquired to minimize the imaging time.

We assessed 35 images with 105 ROIs in T2-weighted MRI that included 21 ROIs for the genu, 63 for the body, and 21 for the splenium. Of the 35 images, 21 were corresponded to the MTR maps and that identified the body and splenium only, due to differences in imaging protocols (Figures 3). We found that there was no significant difference between left, central, and right ROIs of the corpus callosum using any method (p>0.05). While MTR showed a trend to difference between the body and splenium, it did not reach significance and there was no difference between anatomical locations of the corpus callosum in texture heterogeneity. Between animals, texture heterogeneity exhibited the smallest variance between animals, with the mean and standard deviation (bracket) ranged from 0.176 (0.06) to 0.192 (0.05). The range of MTR was intermediate, from 0.473 (0.04) to 0.531 (0.02), and MWF had the largest variability, ranging from 0.403 (0.02) to 0.913 (0.02) (Figure 4). There were no significant outliers in any of these measurements or in any animal.Changes in myelin integrity are associated with many neurological diseases that demand reliable assessment. MTR and MWF are two of the most recognized measures of myelin content, but both need additional pulse sequences for acquisition. Based on clinical scans, texture heterogeneity measures the organization of tissue structure indicated by the distribution pattern (texture) of MRI voxels. Previously, studies showed that texture heterogeneity derived from T2-weighted MRI correlated with the severity of pathology in patients with multiple sclerosis, and that the relationship was stronger with myelin content than with axonal injury and inflammation. In this study, we found that texture heterogeneity showed a similar trend as MTR but with less variability between animals, consistent with the uniformity of myelin as expected in healthy white matter. MWF showed a large variability between animals, likely due to the small number of MRI slices or the quality of multi-echo images acquired that require further confirmation.All three methods confirmed the uniformity of myelin structure in the corpus callosum between hemispheres. Given the advantage of having high consistency between animals, texture heterogeneity deserves further verification as an alternative measure of myelin integrity using standard MRI scans.