Correlation of DTI Metrics to Spinal Cord Cross Sectional Area in Pediatric Subjects with Spinal Cord Injury
Devon M Middleton1, Shiva Shahrampour1, Scott H Faro1, Sona Saksena2, Mahdi Alizadeh1, Chris J Conklin2, Winston Liu3, Govind Nair4, Laura Krisa2, MJ Mulcahey2, and Feroze B Mohamed2

1Temple University, Philadelphia, PA, United States, 2Thomas Jefferson University, Philadelphia, PA, United States, 3University of Maryland, College Park, MD, United States, 4National Institutes of Health, Bethesda, MD, United States

Synopsis

As more advanced imaging techniques are used for the spinal cord (such as DTI) it is important to examine potential relationships to other injury biomarkers. This study was to examine correlations between DTI metrics and spinal cord cross sectional area in pediatric subjects with spinal cord injury. Cord cross section data was acquired using a recently developed technique for semi-automated spinal cord segmentation and measurement of spinal cord cross sectional area. Fractional anisotropy was found to be strongly significantly correlated with spinal cord cross sectional area in the injured subjects.

Purpose

The injured spinal cord undergoes considerable changes, both morphologically and microstructurally from the moment of injury through the acute and chronic stages. As more advanced imaging techniques are used for the spinal cord (such as DTI) it is important to examine potential relationships to other injury biomarkers. The goal of this study was to examine correlations between DTI metrics and spinal cord cross sectional area (SCCSA) in pediatric subjects with spinal cord injury (SCI). SCCSA data was acquired using a newly developed technique for semi-automated spinal cord segmentation and measurement of spinal cord cross sectional area.

Methods

Imaging

Nine pediatric subjects (mean age 11.8) with chronic SCI were scanned using a 3T Siemens Verio MR scanner. Inclusion criteria were: stable cervical or thoracic injury at least 6 months post-injury, and no metal instrumentation or stabilization hardware. All subjects and parents provided informed consent and assent for the IRB approved protocol. Subjects were scanned with an inner-FOV EPI DTI sequence with 2DRF excitation pulses(1). DTI images (Figure 1a) were acquired in two axial segments covering the entire cord with parameters: FOV = 164 x 47 mm; voxel size = 0.8 x 0.8 x 6 mm3, slices = 40, 3 averages of 20 diffusion directions, 6 b0 acquisitions, b = 800 s/mm2, TE = 110 ms, TR = 7900 ms, acquisition time = 8:49 min. Matching axial T2 weighted GRE images were also acquired for anatomic localization. For SCCSA measurement, subjects were scanned with a sagittal 3D TSE T2 weighted isotropic sequence (Figure 1b) with parameters: FOV = 256 x 256 mm, voxel size = 1 x 1 x 1 mm3, TE = 122 ms, TR = 1500 ms, acquisition time = 3:21. Two acquisitions were obtained for each subject with one covering the cervical to upper thoracic cord, and a second covering upper to lower thoracic. Total coverage was C1 through the T12-L1 disc with at least one vertebral level of overlap to ensure effective stitching of the two slabs.

SCCSA Measurements

Images from both 3D isotropic acquisitions were stitched together using FSL(www.fmrib.ox.ac.uk/fsl/) in order to create a single contiguous volume containing the entire cervical and thoracic spinal cord. SCCSA measurement was performed using a newly developed semi-automated segmentation and measurement technique where contours of the spinal cord are identified and automatic segmentation and cross section measurement is preformed in the axial plane(2). SCCSA was averaged for each vertebral level to account for potential differences in subject height/cord length.

DTI

DTI images were corrected for motion (3) and tensor estimation was performed using a non-linear implementation of the RESTORE algorithm (4). Whole cord ROIs were drawn manually on axial images for vertebral levels identified by a board certified neuroradiologist with a sparing of approximately one voxel at the edge to reduce partial volume averaging with CSF. DTI metrics FA, MD, AD, and RD were then calculated from the fitted diffusion tensors.

Results and Discussion

DTI parameters and SCCSA were successfully calculated for all subjects (Figure 2) and correlations were examined between averaged values for the full cord. A strong and statistically significant correlation was found between FA and SCCSA, with Spearman’s r = 0.75 and p < 0.02. Moderate correlations were found with AD (r = 0.50) and RD (r = -0.58) but neither were found to be significant with p = 0.12 and p = 0.07, respectively.

Spinal cord volume loss due to atrophy is expected in SCI, but it is not possible to infer the condition of remaining white matter from conventional structural imaging. While it could be possible that normal appearing tissue is functionally unaffected, the correlation to decreased FA along with atrophy suggests that this is not necessarily the case as the remaining tissue itself shows decreased directionality along with loss of volume. These results suggest that DTI, and potentially other functional imaging, may be a critical compliment to conventional methods when assessing damage to the spinal cord.

Conclusion

The methods used were successful in acquiring DTI metrics and SCCSA for subjects with spinal cord injury. FA was found to be strongly significantly correlated to SCCSA and AD/RD were moderately correlated but not significant. Further examination of the whole cord, as well as level by level comparison, with combined morphological and functional data in spinal cord injury is important as a future path for investigation.

Acknowledgements

This work was funded by the National Institutes of Health (Grant #R01 NS079635-01A1)

References

1. Finsterbusch J. High-resolution diffusion tensor imaging with inner field-of-view EPI. JMRI 2009.

2. Liu W, et al. In vivo imaging of spinal cord atrophy in neuroinflammatory diseases. Ann Neurology 2014.

3. Middleton D, et al. Investigation of Motion Correction Algorithms for Pediatric Spinal Cord DTI in Normals and Patients with SCI. MRI 2014.

4. Chang L, et al. RESTORE: Robust estimation of tensors by outlier rejection. MRM 2005.

Figures

Representative images for midline sagittal of upper slab for 3D TSE acquisition (a) and axial FA map for DTI acquisition

Average whole cord SCCSA and DTI metrics for all subjects



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4407