Quantitative cardiac T1 mapping involves nonlinear parameter estimation after MR acquisition. The widely used minimum mean square error (MMSE) estimator assumes Gaussian additive noise, and can be sensitive to outliers of non-Gaussian nature, such as those caused by cardiac motion. In this work, we propose to apply robust loss functions, which are part of the M-estimator family, with increased robustness to outliers. Experiments on MOLLI and SAPPHIRE sequences showed that the M-estimators were able to improve the T1 estimation robustness, significantly reducing the standard deviation (SD) error of the estimated T1 map in comparison to MMSE.
This abstract and the presentation materials are available to members only; a login is required.