A framework for determining regions of significant anisotropy in whole body diffusion tensor imaging. A simulation and patient study.

Motivation: Reporting the uncertainty of quantitative parameters from MRI studies, such as fractional anisotropy, can significantly enhance the radiological interpretation of derived images. Current techniques lack metrics to indicate uncertainty of FA measures. Goal(s): Difference in Bayesian Information Criterion as a fast method to visualise and compare, the likelihoods of isotropic versus anisotropic models of diffusivity. Approach: The BIC method is validated in a simulation study that probes biologically-relevant FA values outside the brain, and in real patients with metastatic prostate and breast cancer who underwent whole-body DTI. Results: ΔBIC metric is robust to determine where FA is present in whole-body DWI exams. Impact: This method shows the robustness of ΔBIC to indicate significant diffusion anisotropy at each anatomical location.