An optimized MP2RAGE protocol for spinal cord imaging in multiple sclerosis

BACKGROUND AND PURPOSE: MRI methods for assessing spinal cord atrophy remain underutilized in MS research and clinical practice, because even the current advanced acquisition approaches to gray matter and total cord atrophy are hampered by the presence of lesions and inefficiencies in scan time. However, improved acquisition strategies and quantitative assessment methods have the potential to provide atrophy estimates sufficiently precise to be used as outcomes in clinical trials, as well as for patient care, while simultaneously providing excellent lesion identification. We developed and optimized a novel acquisition strategy for the MP2RAGE sequence at 3T to achieve a short acquisition-time protocol suitable for simultaneous estimation of gray and white matter spinal cord areas and lesion detection. MATERIALS AND METHODS: After MP2RAGE protocol optimization using T1 mapping in a pilot cohort of 23 MS participants with upper cord lesions, we applied the optimized protocol to a cohort of 35 MS participants with varying extent of spinal cord involvement. Two neuroradiologists independently detected lesions using the MP2RAGE UNI image and an axial MEDIC sequence routinely used clinically. Inter-rater reliability of the reads was assessed, and generalized mixed models and Bowker's/McNemar's tests were used to compare the lesion-detection efficacy of the two protocols. SNR and CNR of the MP2RAGE images were computed at three representative cervical levels. RESULTS: The proposed protocol provides multiple contrasts that enable clear differentiation of gray matter/white matter and cord/cerebrospinal fluid, even in the presence of lesions - features essential for reliable atrophy measurements. The lesion-detection efficacy of the MP2RAGE was statistically significantly superior to the MEDIC, for both number of detected lesioned levels and reader's confidence. At all three representative levels, the CNR of the optimized protocol was comparable to what was measured for established protocols in a previous study. CONCLUSIONS: We present a 3T spinal cord-optimized MP2RAGE protocol capable of covering the entire cervical spinal cord in approximately 5 minutes. This protocol offers high tissue CNR-fundamental for area quantification-, as well as lesion-detection efficacy comparable to that of established methods. Therefore, such a single MRI protocol could be an efficient approach for conducting clinical research.