Abstract
Quantitative susceptibility mapping (QSM) is an advanced MRI technique that links phase variations to the local tissue susceptibility. In multiple sclerosis (MS), QSM has shown promise in characterizing brain lesions by assessing chronic inflammation and myelin content. However, in the spinal cord (SC), the question remains as to whether QSM can classify MS lesions as demyenilated/remyenilated and detect chronic inflammation. SC QSM poses novel challenges due to the small size, mobility and curvature of the cord, as well as magnetic field inhomogeneities caused by nearby vertebrae and physiological movement. This study compares two QSM processing methods for the SC. One method uses the IDEAL (iterative decomposition of water and fat with echo asymmetry and least-squares estimation) algorithm to account for potential fat-related signal contributions. The second assumes fat has negligible impact. Both approaches employ the PDF (projection into dipole fields) algorithm to remove the background field and the MEDI (morphology enabled dipole inversion) technique for solving the field-to-susceptibility inversion. A 2-sequence MRI protocol was developed to acquire in-phase (IP) and out-of-phase (OOP) QSM data. Eight healthy controls (HC) and twenty MS patients were scanned on a 3T Prisma scanner. We obtained the first high-resolution axial QSM maps of the cervical SC (C3-C5), clearly distinguishing gray matter (GM), white matter (WM), and MS lesions. Results showed that accounting for fat using IDEAL did not meaningfully improve the estimation of the total magnetic field or the overall quality of QSM maps. These results open up exciting possibilities for SC susceptibility imaging in MS.