Synopsis

A 1-dimensional MR sequence termed Fine Structural Analysis (fineSA) was applied to the human cervical spinal cord in order to determine if spinal tracts and nerve fibres of the dorsal horn could be identified from the spectral regularity of their myeloarchitecture. Repeatable 60 μm peaks could be attributed to the dorsal horn as opposed to projective nerve tracts of the cord. This analysis might be able to identify the loss of neurons via changes in packing density that would be proximal to macrostructural atrophy detectable on more classical MR sequences.

Introduction

A 1-dimensional MR sequence termed Fine Structural Analysis (fineSA) has been developed and demonstrated for trabecular bone pore analysis¹ that was able to identify feature sizes down to 56μm. In cortical grey matter a resolution down to 38μm was demonstrated². In this study, the fineSA technique was applied to the human cervical spinal cord in order to determine if spinal tracts and nerve fibres of the dorsal horn could be identified. If possible, this analysis might be able to identify the loss of neurons and their projective axons resulting in a loss of packing density that would be proximal to macrostructural atrophy detectable on more classical MR sequences.

Materials and Methods

The fineSA method uses an orthogonal 90-180 pair of slice-selective pulses to excite a “prism” that is read out with a strong readout gradient. This produces a 1-dimensional spectrum that can be decomposed into component frequencies.

The initial validation phantom used 8 μm ultra-high-molecular-weight polyethylene (UHMWPE) (Dyneema, DSM Heerlen, NL) fibres wound around a former, as used for diffusion kurtosis verification³ mimicking white matter bundles. UHMWPE has little susceptibility shift versus water, but it was not possible to achieve a regular physical packing (Figure 1). As an alternative, a plastic 3D microgrid was designed and manufactured in house (Figure 2) using a 3D printer modified to print small objects at much higher resolution with a custom print head that can print high temperature nylon.

Phantom and human cervical cord scans were acquired on a Siemens 3T Tim Trio and Verio (Siemens Healthcare, Erlangen) using a 10cm loop coil placed behind the neck.

Phantom prisms were 1x1x10cm, with 4096 readout points in 48ms, giving 0.0244mm resolution along the prism with 21Hz/pixel bandwidth. The in-vivo prism used was 1x1x5cm, with 2048 readout points for the same resolution as phantom scans. 256 averages were acquired with TR=1s, requiring 4m:16s per scan location. Prisms were placed axially at the C3 and C4 level, obliqued to place the long axis of the prism nominally normal to the right dorsal horn. An additional prism was acquired with the long axis of the prism along the spinal cord, nominally covering C5 up towards C2.

Results

Difficulty in attaining a regular packing of the 8μm fibres limited interpretability of the fineSA scans in the diffusion phantom, where multiple regularities were identified from 80 to 300 μm depending on the pressure applied by the former plates. The coarser 3D printed structure was readily and strongly identified at the correct feature sizes.

The in vivo scans found a strong and reproducible signal at 60 μm between C3 and C5. The dorsal horn at C3-C5 is characterized by anatomically well-defined crossing bands of nerve fibres. Thus the scans show expected results for myeloarchitectural features associated the nerve root entry in the dorsal horn. Consistent with assignment to myelin at root entry zones, this periodicity was found in the dorsal horn but not elsewhere across regions interrogated with the prism. In one additional subject (Figure 4), the spatial frequency was checked with a vertically positioned prism. Nerve fibres running along the cord would not be detected in this position, only the axially running nerve roots.

Discussion

In this study we demonstrated a repeatable signal at 60 μm periodicity in the spinal cord at the C3 to C5 levels. The signal was attributed to the dorsal nerve roots rather than nerve tracts of the cord. We detected periodicities of appropriate dimensions reproducibly requiring only ~5 min per prism. The dimensions achieved are nearly an order of magnitude below that which can be achieved with conventional MR imaging, even at ultra-high field strengths such as 7T.

Acknowledgements

We gratefully acknowledge Acuitas Medical systems for providing the FineSA sequence and analysis software, and the MS Society for funding support.

References

1. Evans BAJ, James TW, James K, Cox A, Farr L, Paisey SJ, et al. Preclinical Assessment of a New Magnetic Resonance-based Technique for Determining Bone Quality by Characterization of Trabecular Microarchitecture. Calcif Tissue Int. 2014 Dec 8;95(6):506–20.

2. Kristen James, S.C., Tim James, Lance Farr, James Raffety, Gareth Thomas MB, David Chase, Christopher Rogers, Peter Jezzard SG. A new magnetic resonance based approach to assessment of pathology in early Alzheimer’s disease. In: 7th Clinical Trials Conference on Alzheimer’s Disease. 2012.

3. Farrher E, Kaffanke J, Celik AA, Stöcker T, Grinberg F, Shah NJ. Novel multisection design of anisotropic diffusion phantoms. Magn Reson Imaging. Elsevier Inc.; 2012;30(4):518–26.