A Multi-Variate framework to assess reliability and discrimination power of Bayesian estimation of Intravoxel...

Area of Science:

• Medical Imaging
• Diffusion MRI

Background:

• Intravoxel Incoherent Motion (IVIM) imaging provides insights into tissue microstructure.
• Accurate estimation of IVIM parameters is crucial for clinical applications, particularly in oncology.
• Assessing the reliability and discriminative power of IVIM models is essential for robust interpretation.

Purpose of the Study:

• To propose a multivariate, multi-step framework for systematically evaluating the estimation reliability and discriminability of IVIM model parameters.
• To establish guidelines for interpreting IVIM data in clinical settings.

Main Methods:

• Monte-Carlo simulations were performed with varying signal-to-noise ratios (SNRs) and IVIM parameter combinations.
• Simulations utilized two b-value discretizations (24 and 9 values) and a Bayesian fitting approach.
• The framework assessed model selection (mono- vs. bi-exponential using BIC), fitting accuracy, and the ability to discriminate between different IVIM parameter distributions using multivariate tests.

Main Results:

• Bi-exponential fitting demonstrated reliability for perfusion fraction >5%, with high accuracy for D and acceptable error for f, but significant uncertainty in D*.
• Discrimination between two parameter distributions is feasible with differences in D (≥0.3 x10⁻³ mm²/s).
• Discriminating similar D values requires a 5% difference in f, contingent on balanced sample size and dense b-value discretization; D* had negligible impact.

Conclusions:

• The IVIM model offers accurate diffusion estimation, but perfusion parameter uncertainties can limit its discriminative power.
• The proposed framework provides interpretative guidelines for reliable IVIM data analysis.
• Adopting this framework aids in understanding the limitations and strengths of IVIM in differentiating patient populations.