A robust framework for characterising diffusion metrics of the median and ulnar nerves: Exploiting state‐of‐the‐art tracking methods

Abstract

Diffusion-weighted imaging has been used to quantify peripheral nerve properties; however, traditional post-processing techniques have several limitations. Advanced neuroimaging techniques, which overcome many of these limitations, have not been applied to peripheral nerves. Here, we use state-of-the-art diffusion analysis tools to reconstruct the median and ulnar nerves and quantify their diffusion properties.Diffusion-weighted MRI scans were obtained from eight healthy adult subjects. Constrained spherical deconvolution was combined with probabilistic fibre tracking to compute track-weighted fibre orientation distribution (TW-FOD). The tensor was computed and used along with the tracks to estimate TW apparent diffusion coefficient (TW-ADC), TW fractional anisotropy (TW-FA), TW axial diffusivity (TW-AD), and TW radial diffusivity (TW-RD). Variability of TW measurements was used to estimate power size information. The population intersession mean (± SD) measurements for the median nerve were TW-FOD 1.30 (±0.17), TW-ADC 1.16 (±0.13) Â 10 À3 mm 2 /s, TW-FA 0.60 (±0.05), TW-AD 2.05 (±0.16) Â 10 À3 mm 2 /s, and TW-RD 0.72 (±0.12) Â 10 À3 mm 2 /s. The corresponding measurements for the ulnar nerve were TW-FOD 1.25 (±0.14), TW-ADC 1.13 (±0.10) Â 10 À3 mm 2 /s, TW-FA 0.56 (±0.06), TW-AD 1.93 (±0.01) Â 10 À3 mm 2 /s, and TW-RD 0.74 (±0.12) Â 10 À3 mm 2 /s. Based on these measurements, a sample size of 37 would be sufficient to detect a 10% difference in any of the measured TW metrics. A sample size of 20 would be large enough to detect within-subject differences as small as 2.9% (TW-AD, ulnar nerve) and between-subject differences as small as 3.8% (TW-AD, ulnar nerve).

Meta-analysis of the normal diffusion tensor imaging values of the peripheral nerves in the upper limb

“…Studies were performed most commonly on Siemens (9 studies 31,[38][39][40][41][42][43] , 43%) or Philips (9 studies 29,44,[44][45][46][47][48][49][50][51] , 43%) scanners with the remainder using GE (3 studies 28,52,53 , 14%). Nineteen studies were performed at 3T 28,29,[38][39][40][41][42][43][44][46][47][48][49][50][51][52][53][54] , with one at 1.5T 45 and another at 7T 31 . The majority of studies used single-shot echo planar imaging (ssEPI; 17 studies, 85%), one compared ssEPI to readout-segmented echo planar imaging (rsEPI) 39 and 3 studies 28,31,40 did not specify the type of sequence.…”

“…The majority of studies used single-shot echo planar imaging (ssEPI; 17 studies, 85%), one compared ssEPI to readout-segmented echo planar imaging (rsEPI) 39 and 3 studies 28,31,40 did not specify the type of sequence. Subjects were most commonly in the "superman" position (i.e., prone with the shoulder and elbow extended, with the elbow positioned in the isocentre of the magnet; 17 studies 28,29,31,38,38,39,[41][42][43]45,46,[48][49][50][51][52]54 , 85%) whilst the others positioned individuals supine with their arm extended overhead 40 , supine in the anatomical position 44,47 or did not report the position 53 . The receiver coils used were most commonly flexible extremity coils (11 studies 29,31,39,[41][42][43][44]47,51,54 , 55%) with the remainder using head 48 , knee 38,46,50,52 , wrist 28,45 or unspecified 53 coils.…”

“…Nine studies used parallel imaging techniques (GRAPPA [39][40][41][42]42 , SENSE 29,29,[49][50][51] , ASSET 52 , and one unspecified method 31 ) whilst 9 studies 28,38,[44][45][46][47][48]53,54 did not specify this parameter. Four studies used partial Fourier acquisition [39][40][41]51 , one reported full k-space acquisition 52 but the majority of studies 28,29,31,38,[42][43][44][45][46][47][48][49][50]53,54 did not report this information. Nineteen studies captured a single (maximum) b-value of mean 1045 s/mm 2 (range 700-1300).…”

“…Nineteen studies captured a single (maximum) b-value of mean 1045 s/mm 2 (range 700-1300). One study 52 captured several b-values (300, 450, 600, 750 and 900 s/mm 2 ) to calculate track-weighted DTI metrics via multi-shell multi-tissue constrained spherical deconvolution. All other studies reconstructed their data using 2 nd order tensors.…”

“…The median N D was 20 (range 6-64). Most studies did not specify the diffusion encoding waveform 28,29,31,[42][43][44][45][46][47][48][49][50][51]53,54 whilst 6 studies [38][39][40][41]52 used "monopolar" with no further explanation.…”