Multicomponent Diffusion-Weighted MRI: Revealing Microenvironment Changes in Degenerative Spinal Cord

Diffusion-weighted Imaging (DWI) is a well-established magnetic resonance imaging technique commonly used in studying neurodegeneration in the brain and spinal cord. The DWI techniques have predominantly focused on brain investigations and the studies on spinal cord share technical similarities with those conducted on the brain. However, there are notable differences between the brain and spinal cord. The brain exhibits greater isotropy, whereas the spinal cord demonstrates more anisotropy. Consequently, specific DWI techniques that take advantage of the spinal cord's anisotropic nature may outperform the commonly used techniques designed for isotropic objects. Unfortunately, research on the application of these specific DWI techniques in the spinal cord is limited. Hence, the aim of this dissertation is to explore this new regime and expand the knowledge of a specialized DWI technique, ultra-high b-value DWI, for the classification of healthy and Amyotrophic Lateral Sclerosis (ALS) affected mouse spinal cords. To address the knowledge gap surrounding the utilization of ultra-high b-value techniques, a systematic study incorporating simulation, phantom, ex vivo, and in vivo investigations was conducted. The dissertation begins with simulation and phantom studies, which validated the feasibility of employing regularized (Non-negative Least Squares) NNLS methods for decomposing DWI data. Subsequently, the ex vivo study demonstrated the potential to encode diffusion contrast from ultra-high b-values onto DW images. Moreover, the ex vivo data was analyzed using the well-established L2-norm regularized NNLS method in Myelin Water Imaging, successfully capturing microenvironmental changes in mouse spinal cords caused by ALS. However, the L2-norm regularization exhibited an over-smoothing effect in the results. To address this issue, a novel L1-norm regularized NNLS method was introduced and applied to analyze the same ex vivo datasets. Additionally, the decomposition nature of NNLS methods facilitated another study that explored the impact of low b-values on DW signals. Drawing upon the knowledge and insights gained from the preceding studies, DWI protocols incorporating ultra-high b-values were developed for in vivo mouse spinal cord imaging. The results obtained using the novel L1-norm regularized NNLS method on the in vivo data unveiled new imaging indices for classifying ventral roots in wild-type and ALS-affected animals at the pre-symptomatic stage (P75) and early symptomatic stage (P90). Furthermore, lower signal intensities were observed in the ventral roots of ALS-affected spinal cords, and the results obtained from the proposed L1-norm regularized NNLS method affirmed a shift towards higher diffusion coefficient ranges in weight distributions. These findings may potentially describe changes in the local environment as the disease progresses.