Cramér-Rao bound for Intravoxel Incoherent Motion Diffusion Weighted Imaging fitting.

ABSTRACT

The precision of parameter estimation for Intravoxel Incoherent Motion Diffusion Weighted Imaging (IVIM-DWI) was investigated by examining their Cramér-Rao bounds (CRBs) under the presence of Rician noise. Monte Carlo (MC) simulation was also conducted to validate the CRB results. The estimation uncertainties of true diffusion coefficient (D) and perfusion fraction (f0) could reach 3.89% and 11.65% respectively with typical parameter values at a moderate signal-to-noise ratio (SNR) of 40. However, to estimate pseudo diffusion coefficient (D*) within 10% uncertainty requires SNR>122. The results also showed that the estimation precision of each parameter is not only dependent on SNR but also their true values, while this mutual dependency is complicated. Under some particular cases, estimation uncertainty for certain parameters might be smaller than 5% at a moderate SNR of 40. However, the simultaneous precise estimation for all three parameters is theoretically difficult and highly SNR demanding.