Characterization of a spinal cord diffusion tensor imaging pipeline with pathological spine data

Abstract

Fiber tractography from magnetic resonance (MR) diffusion tensor imaging (DTI) enables the visualization of white matter bundles. In the presence of pathology, these bundles can be distorted and disconnected, which can reveal clinically significant information about the nature of the underlying pathology. This work studies DTI in the spine in the presence of pathology. A spine DTI pipeline that was developed in an earlier study is evaluated against the pathological data. We study the challenges of adapting the pipeline to pathological spine data, where MRI artifacts and significant distortion in cord shape and contrast from pathology make automated cord segmentation and registration extremely challenging. Moreover, we identify challenges with processing highly anisotropic MRI volumes and the implications this has on DTI processing. Heuristics are developed to handle these issues and are incorporated into the pipeline. Finally, visualizations of the tractography streamlines are generated and the impact of pathology on the streamline trajectories is briefly discussed, awaiting clinical validation.