Improved MRI quantification of spinal cord atrophy in multiple sclerosis.

PURPOSE: To identify an improved method for measuring spinal cord cross-sectional area (CSA) using magnetic resonance imaging (MRI) in multiple sclerosis (MS). MATERIALS AND METHODS: MRI was performed on 15 controls and 15 MS patients and repeated in nine controls and nine patients after 6 months. At this timepoint, an additional scan was acquired to evaluate scan-rescan reproducibility. Two sequences were acquired in the cervical cord: 3D phase sensitive inversion recovery (PSIR) and 3D magnetization prepared rapid acquisition T1-weighted gradient echo. CSA was outlined at C2-C3 using two methods: a semiautomated edge detection method and active surface model (ASM). We evaluated reproducibility for all combinations of sequences and analysis methods using coefficient of variation (COV) and intraclass correlation coefficient and performed sample size calculations for clinical trials to reduce longitudinal cord atrophy. RESULTS: PSIR/ASM combination provided the lowest values of COV for intrarater, interrater, scan-rescan reproducibility (0.002%, 0.03%, and 0.1% respectively). At 6-month follow-up no significant changes were seen in CSA of controls, and a trend towards significance was observed in patients. CONCLUSION: PSIR/ASM proved more reproducible than established methods of evaluating CSA in MS and also provides the lowest number of subjects per arm for 6-month and 1-year clinical trials.