Region-specific analysis of diffusion tensor imaging for cervical spondylotic myelopathy

Abstract

Cervical Spondylotic Myelopathy (CSM) is a common type of spinal cord dysfunction in the elderly. The natural history of CSM is associated with disc degeneration and spondylosis, leading to the static and dynamic compression of the spinal cord, tissue ischemia, tissue damage, and ultimately neurological function deficit. However, the severity of the spinal cord compression does not necessarily correlate with the signs and symptoms of CSM in patients. Until now, the pathomechanism of CSM was not well understood. Establishing an evaluation technique is, therefore, criticalfor the pathophysiological investigation of CSM.

Magnetic resonance imaging (MRI) has been widely used for evaluating the spinal cord parenchyma. However, conventional MRI is limited in detecting macroscopic changes, e.g. spinal cord compression, edema or hemorrhage etc. Recently, there has been increasing interest in diffusion tensor imaging (DTI), which permitting detects tissue water molecule diffusion at the microscopic level.

The conventional DTI analysis for CSM relies on hand-drawn regions of interest (ROIs), so called ROI-based measurements. The ROIs are drawn on the sagittal image or on the axial image to cover the whole cord, which are insufficient to describe the precise diffusion pattern. In particular, the deformation and degeneration of the myelopathic cord poses a big challenge for the ROI-based analysis. The most commonly used parameter, fractional anisotropy (FA) has difficulty in determining the level diagnosis due to its relatively large variance along the cord. Furthermore, the functional activation following microstructural damage remains underexplored.

In this dissertation, several novel methods for region-specific analysis were proposed for the investigation of microstructural changes in the CSM. In Chapter 2, ROI-based analysis was employed to detect the regional diffusion characteristics in CSM. In Chapter 3, an auto-template was developed that segments the cord and measures the DTI parameters automatically. We found that our auto-template outperforms hand-drawn ROI-based methods in terms of efficiency and reproducibility. In Chapter 4, entropy-based analysis was proposed to characterize the loss of complexity of microstructure in the myelopathic cord. It was demonstrated that FA entropy was an objective and quantitative evaluation parameter that was superior to conventional methods for separating CSM patients from healthy subjects. In Chapter 5, orientation entropy was used to detect the disordered orientational distribution of the nerve tracts in CSM, which could be used as a good index for the pathogenic level estimation. In Chapter 6, a diffusion tensor tractography-based method was proposed to overcome the difficulties in column-specific ROI drawing on the deformed and degenerated spinal cords. In Chapter 7, the structure-function relationship in the cervical spinal cord was explored by a combination of DTI and functional MRI. A significant correlation was found between enhanced functional responses and the loss of microstructural integrity in CSM.

In this study, several novel post-processing methods were proposed and demonstrated, which were shown to have extraordinary capabilitiesfor the investigation and assessment of CSM. It is expected that these methods can be used as valuable tools for clinical diagnosis and for the selection of the most appropriate treatment strategy for CSM.