Quantitative magnetization transfer (qMT) imaging provides quantitative information of macromolecular properties of tissues, but requires a long scan time. In this study, a neural network is proposed for the acceleration of qMT imaging. The network was trained to output the full 12 MT images (6 offset frequencies, 2 RF powers) from an input of only 4 MT images (2 offset frequencies, 2 RF powers). The qMT imaging with the neural network showed results comparable to those from the conventional qMT imaging with the full 12 MT images, indicating reduction in scan time by a factor of 3.
