Quantitative Nerve Microstructure Analysis

The nerve segments of peripheral nerves, acquired in 16 continuous slices of 0.625 mm thickness, were identified on reference T₂-weighted images (Figure 1A). Quality assessment of slices was performed, and slices with artifacts and partial volume effect were excluded from further analysis. In each included slice, fascicles, interfascicular epineurium, perineurium, and nerve cross-sectional area (CSA) were segmented. The fascicles were defined as intraneural hypointense oval- or round-shaped tissue circumferentially surrounded by a markedly hyperintense line representing the perineurium. The latter served as a reliable segmentation border (Figure 1B). The perineurium was segmented with a single measurement by two parallel lines, as shown in Figure 1C. The hyperintense tissue between the fascicles was defined as interfascicular epineurium (Figure 1D). The nerve was segmented to include the entire nerve but a minimal proportion of the background (Figure 1E). Segmentations were performed manually with the image processing software ImageJ (National Institutes of Health, Bethesda, Maryland, United States). The area was recorded for each segmentation and expressed as CSA for the nerve and fascicles. The fascicular ratio (FR) was calculated as a net fascicular CSA/nerve CSA, the ratio of perineurium as a net perineurium/nerve CSA, and the ratio of interfascicular epineurium as net interfascicular epineurium/nerve CSA (Tagliafico and Tagliafico, 2014 (link)).
The diffusion tensor was calculated from the acquired three-dimensional data as described previously (Basser et al., 1994 (link); Awais et al., 2022 (link)). For each image voxel, the calculated diffusion tensor was diagonalized, which yielded maps of the tensor eigenvalues D₁, D₂, and D₃ and of the corresponding eigenvectors ( ${\stackrel{\rightharpoonup }{\mathrm{\epsilon }}}_1,{\stackrel{\rightharpoonup }{\mathrm{\epsilon }}}_2$ , ${\stackrel{\rightharpoonup }{\mathrm{\epsilon }}}_3$ ). Diffusion tensor and its diagonalization were also calculated for every delineated compartment from the corresponding average diffusion weighted signals of the compartment for 19 different diffusion gradient directions. Regional signal averaging enabled the calculation of the fractional anisotropy (FA), mean diffusivity (MD), and D_||/D_⊥ using the equations Eqs. 1–3 with less noise for each of the segmented compartments. The calculations were made using the software written in the C programming language, which has been previously developed and specifically modified by the authors (Awais et al., 2022 (link)). $\mathrm{M}\mathrm{D}=\frac{D_1+D_2+D_3}{3}$
$\mathrm{F}\mathrm{A}=\sqrt{\frac{3}{2}}\frac{\sqrt{{\left(D_1-\mathrm{M}\mathrm{D}\right)}^2+{\left(D_2-\mathrm{M}\mathrm{D}\right)}^2+{\left(D_3-\mathrm{M}\mathrm{D}\right)}^2}}{\sqrt{D_1^2+D_2^2+D_3^2}}$
$D_{\Vert }/D_{\perp }=\frac{D_1}{\frac{D_2+D_3}{2}}$