Theoretical and experimental evaluation of broadband decoupling techniques for in vivo nuclear magnetic resonance spectroscopy.

A theoretical and experimental evaluation of existing broadband decoupling methods with respect to their utility for in vivo (1)H-(13)C NMR spectroscopy is presented. Simulations are based on a modified product operator formalism, while an experimental evaluation is performed on in vitro samples and human leg and rat brain in vivo. The performance of broadband decoupling methods was evaluated with respect to the required peak and average RF powers, decoupling bandwidth, decoupling side bands, heteronuclear scalar coupling constant, and sensitivity toward B(2) inhomogeneity. In human applications only the WALTZ and MLEV decoupling methods provide adequate decoupling performance at RF power levels that satisfy the FDA guidelines on local tissue heating. For very low RF power levels (B(2max) < 300 Hz) one should verify empirically whether the experiment will benefit from broadband decoupling. At higher RF power levels acceptable for animal studies additional decoupling techniques become available and provide superior performance. Since the average RF power of adiabatic RF pulses is almost always significantly lower than the peak RF power, it can be stated that for average RF powers suitable for animal studies it is always possible to design an adiabatic decoupling scheme that outperforms all other schemes. B(2) inhomogeneity degrades the decoupling performance of all methods, but the decoupling bandwidths for WALTZ-16 and especially adiabatic methods are still satisfactory for useful in vivo decoupling with a surface coil.