Modeling pulsed magnetization transfer

Abstract

Modeling the effects of clinical magnetization transfer (MT) scans, which generate contrast using short, shaped radiofrequency (RF) pulses (pulsed MT), is complex and time-consuming. As a result, several studies have proposed approximate methods for a simplified analysis of the experimental data. However, potential differences in the MT parameters estimated by each method may complicate the comparison of reported results. In this study we evaluated three approximate methods currently used in quantitative MT (qMT) studies. In the first part of the investigation, an MT modeling technique that makes minimal approximations, other than the use of a two-pool tissue representation, was developed and validated. Subsequently, this technique served as a standard against which to evaluate the other, more approximate models. Each model was used to fit experimental data from samples of wild-type (WT) and shiverer mouse spinal cord, as well as simulated data generated by our minimal approximation modeling technique. The results of this study demonstrate that the approximations used in pulsed MT modeling are quite robust. In particular, it was shown that the semisolid pool fraction, M(0)(B), which is known to correlate strongly with myelin content, and the transverse relaxation time of macromolecular protons, T(2)(B), could be evaluated with reasonable accuracy regardless of the model used.