The purpose of this study was to propose a robust technique for the determination of unbiased kurtosis values in body applications; for this aim, a new strategy to pixelwise estimate the noise level in DKI acquisitions is described and its feasibility for pelvic DKI is demonstrated in prostate tissue.
The "naive" evaluation (without considering the image noise) resulted in significantly biased positive kurtosis values of urine of $$$K=0.42\,($$$standard deviation: $$$0.06)$$$, indicating non-monoexponential signal decay caused by the influence of noise at higher b-values. This bias is almost completely removed ($$$K=-0.01\,(0.32)$$$) after including the pre-calculated noise-level map in the evaluation.
This abstract and the presentation materials are available to members only; a login is required.