Experimental Errors as Alternative to the Theoretical Cramr-Rao Minimum Variance Bounds in MRS: ER-ARSOS - Error Estimation by Multiple Quantification of Recombined ARSOS-Filtered Output Signals

Johannes Slotboom¹, Dirk van Ormondt², Olivier Scheidegger¹, Caspar Brekenfeld¹, Roland Wiest¹, Gerhard Schroth¹, Danielle Graveron-Demilly³

The Cramer-Rao minimum variance bounds (CR-MVB) are widely used as estimates of errors in spectral-parameters in localized MRS. Due to the macromolecular baseline, non-analytical line-shapes, and artefacts resulting from non-stationary signal acquisition conditions or RF-pulse imperfections, the parametric model is normally incomplete, and, strictly taken, the CR-MVBs estimates invalid. In general the CR-MVB under-estimates the variances in the estimated spectral-parameters. This work presents a novel method named ER-ARSOS to estimate the experimental errors of the parameters, as an alternative to the CR-MVBs in MRS.