Fast and accurate parameter estimation of noisy complex exponentials with use of prior knowledge

In this paper we address the problem of parameter estimation of magnetic resonance spectroscopy (MRS) signals. MRS signals are modeled as complex exponentials in noise. Iterative methods based on an optimisation procedure can be used for the parameter estimation. We examine which functional we have to minimise and which nonlinear least squares algorithms we have to use in order to attain maximum efficiency and robustness. The influence of starting values and prior knowledge is examined.