Is spending extra scan time on measuring amacromolecules-only' signal worthwhile?

INTRODUCTION Quotation of reliable error bars on estimated metabolite concentrations is imperative in both research and clinical environments. Yet, estimation of errors is notoriously difficult, especially when derived from `once-only' measurements. Usually, one quotes Cramer-Rao Bounds (CRBs) as surrogates of the errors. These CRBs are necessarily based on estimated model parameters rather than on (a priori unknown) exact model parameters, and may therefore not be sufficiently reliable. The situation is aggravated when the metabolite signal is perturbed by a macromolecule signal. This situation is called semi-parametric, because the MRS signal contains a parametric part -metabolites, model function supposedly known -and a non-parametric part -macromolecules, model function unknown. Whenever it occurs, the error bars of the metabolite concentrations become even less reliable [1-3]. Under the circumstances, one should consider whether it is worthwhile to measure the `macromolecules-only' signal too -by inversion-recovery [4] -and use the latter to render the signal parametric. This in turn could improve the reliability of the error bars of the estimated metabolite concentrations.