Quantitative magnetization transfer (qMT) imaging in skeletal muscle may be confounded by intramuscular adipose components, low signal-to-noise ratios (SNRs), and voluntary and involuntary motion artifacts. Collectively, these issues could create bias and error in parameter fitting. In this study, technical considerations related to these factors were systematically investigated, and solutions were proposed. First, numerical simulations indicate that the presence of an additional fat component significantly underestimates the pool size ratio (F). Therefore, fat-signal suppression (or water-selective excitation) is recommended for qMT imaging of skeletal muscle. Second, to minimize the effect of motion and muscle contraction artifacts in datasets collected with a conventional 14-point sampling scheme, a rapid two-parameter model was adapted from previous studies in the brain and spinal cord. The consecutive pair of sampling points with highest accuracy and precision for estimating F was determined with numerical simulations. Its performance with respect to SNR and incorrect parameter assumptions was systematically evaluated. QMT data fitting was performed in healthy control subjects and polymyositis patients, using both the two- and five-parameter models. The experimental results were consistent with the predictions from the numerical simulations. These data support the use of the two-parameter modeling approach for qMT imaging of skeletal muscle as a means to reduce total imaging time and/or permit additional signal averaging.
Copyright © 2015 Elsevier Inc. All rights reserved.