DTI of the Brain and Cervical Spine: Evaluation in Normal Subjects and Patients With Cervical Spondylotic Myelopathy

More than half of the middle-aged population has radiologic evidence of cervical spondylosis and a subset of this population develops cervical spondylotic myelopathy (CSM), a condition in which the spinal cord is impaired, either by direct mechanical compression or indirectly by arterial deprivation and/or venous stasis. Although many operations are performed for conditions related to CSM, no consensus exists regarding the timing of the surgical intervention or how to select among the available surgical options for a given patient. Many factors have been implicated in the propensity for CSM to develop including advanced age, disability at presentation, cord diameter, cord area, altered cord signal on MRI (T2- and T1- weighted images), increased cervical spinal mobility and a congenitally narrow spinal canal. The same factors may also determine the response to surgery, either positively (increased cervical mobility) or negatively (advanced age, congenitally narrow spinal canal). The potential complications of surgical procedures are noteworthy. Although incapacitating adverse effects are uncommon, their occurrence in a disease process with a potentially benign course is of undoubted gravity. Finally, it should be noted that the radiologic criteria of cervical root or cord compression on MRI are subjective and it is uncertain that they correlate with the clinical symptoms.

Diffusion tensor imaging (DTI) is sensitive to the motion of water and is considered to be a marker for axonal integrity in the brain. There has been much interest in applying the technique to the spine and in particular to the evaluation of patients with CSM. In the brain DTI is performed using a single shot echo planar imaging (SS-EPI) sequence which acquires all of the data for a single slice following the application of a single excitation pulse and diffusion gradients. While very efficient in terms of data collections there are several problems with SS-EPI, notably that 1) It is best suited to the acquisition of relatively small imaging matrices, resulting in either low spatial resolution or the use of additional techniques to reduce the field of view (FOV). Increasing the number of echoes acquired to increase the resolution does not produce an increase in the acquired resolution since the decay of the signal during the longer echo train results in increased blurring of the image. 2) The images are distorted by susceptibility gradients in areas close to the interfaces between tissue and bone or osseous structures. These factors have made applying DTI to clinical studies of the spine problematic. Specifically:-

• The small size of the spinal cord means that high spatial resolution is required.
• The susceptibility effects from the surrounding osseous structure result in image distortion.
• The motion of the cord during the cardiac cycle leads to artifacts.

Readout segmented EPI (RESOLVE or RS-EPI) sequence has been shown to be capable of obtaining images with higher spatial resolution and reduced distortion in the brain. We have recently demonstrated that the RESOLVE sequence can also be used to obtain high quality DTI images of the cervical spine of volunteers (see figure 1) and can provide clinically useful information in subjects with metastases in the spine. The RESOLVE sequence breaks the data collection into several segments, resulting in reduced distortion and the potential to obtain images with higher resolution, but at the cost of either increased scan time or a reduced number of directions for the diffusion gradients (i.e. reduced signal to noise ratio (SNR)). The RESOLVE sequence also incorporates real-time assessment of the data, enabling data corrupted by motion to be rejected and then reacquired. We now wish to optimize, and assess the clinical utility of, the RESOLVE sequence in patients with CSM by applying it to a group of CSM patients and a group of aged matched control subjects. The raw diffusion images are post-processed to yield parametric images which characterize the diffusivity (mean, longitudinal and radial) and anisotropy (directional dependence) of the diffusion. The anisotropy of the diffusion data is typically quantified using the fractional anisotropy (FA) index, however, we have recently shown that two other quantities, the mode and ellipsoidal area ratio (EAR) of the diffusion tensor, provide useful supplementary information and we propose to use all three parameters when analyzing the anisotropy of the diffusion data from this study.

Using the RESOLVE sequence address these first two of the "bullet" points listed above and we found that in some subjects high quality DTI images of the spine can be obtained without the use of cardiac gating, however, for a robust clinical sequence we, and other groups, have found cardiac gating to be essential. Another issue to consider for DTI studies of the spine is the orientation of the scan. Sagittal images provide coverage of the whole spine with relatively few slices, which is particularly beneficial when cardiac gating is employed. Axial slices have the advantage that the changes in the configuration of the spine in the caudal-cranial direction are relatively slow, meaning that thicker slices can be used without a significant loss of information. However, a relatively large number of slices are still required to cover the entire C-spine resulting in a lengthy measurement time when cardiac gating is used.

Objectives The investigators wish to try and answer the following questions for the C-spine:-

• What are the diffusion characteristic of the normal C-spine (spatial variation and age dependence)?
• How do the RESOLVE and SS-EPI images of the spine compare in terms of image quality?
• What is the effect of changing the number of segments for the RESOLVE sequence?
• When comparing SS-EPI and RESOLVE are there statistically significant differences in the derived values of anisotropy or diffusivity?
• Are the diffusion characteristics of the C-spine of subjects with CSM significantly different from those of aged matched controls?
• Can DTI of the spine be used to derive diagnostic information, either in terms of prognostic information for individual CSM patients or the identification of sub-groups of CSM patients? For the brain the question we would be addressing are:-

• When comparing SS-EPI and RESOLVE are there statistically significant differences in the derived values of anisotropy or diffusivity? 2) Is there a statistically significant difference in the volume of the fiber tracts derived using the two sequences or between the two groups? 3) Is the diffusion data from the brains of patients with CSM significantly different from that of the controls?

Inclusion criteria

• 20 patients will be selected by an experienced team of orthopedic specialist from the Emory Orthopaedics & Spine Center and will include patients suffering signs and symptoms of cervical myelopathy and with evidence of radiographic cervical cord compression with or without abnormal spinal cord signal intensity secondary to spondylosis.
• 20 controls : Aged matched to the patient group. No history of CSM.

Exclusion criteria

• Any history of significant trauma implicating the brain or cervical spine, including prior motor vehicle collision and history of prior brain, neck or cervical spine surgery.
• Any history of neurological disease
• Dental braces or similar orthodontic devices.
• Claustrophobia