Optimal Model Mapping for Intravoxel Incoherent Motion MRI

Area of Science:

• Medical Imaging
• Radiology
• Biophysics

Background:

• Intravoxel incoherent motion-magnetic resonance imaging (IVIM-MRI) typically uses a single diffusion model for all voxels.
• The choice of diffusion model significantly impacts diffusion parameter estimation and perfusion analysis reliability.
• Current methods may lack accuracy due to the heterogeneity of diffusion characteristics within different tissues.

Purpose of the Study:

• To propose and evaluate an optimal model mapping method for IVIM-MRI to improve perfusion parameter estimation reliability.
• To compare the performance of the optimal model mapping method against conventional single-model approaches.
• To assess the potential clinical utility of optimal model maps in neurological studies.

Main Methods:

• Six healthy volunteers underwent brain IVIM-MRI using a 3.0 Tesla scanner with 17 b-values (0-2,500 s/mm²).
• Diffusion models (Gaussian, Kurtosis, Gamma) were assessed for optimality in different brain tissues (CSF, WM, GM).
• An optimal model mapping strategy was developed and applied to estimate perfusion fraction (fₚ) and pseudo-diffusion coefficient (D*).

Main Results:

• The Gaussian, Kurtosis, and Gamma models were identified as optimal for CSF, WM, and GM, respectively.
• The optimal model mapping method yielded improved GM/WM ratios for fₚ (1.54) and D* (1.24) compared to single models.
• Estimated fₚ and D* values using the optimal model mapping method demonstrated enhanced reliability over conventional techniques.

Conclusions:

• Optimal model mapping significantly improves the reliability of perfusion parameter estimation in IVIM-MRI.
• This method accounts for tissue-specific diffusion characteristics, leading to more accurate results.
• Optimal model maps offer potential for enhanced clinical diagnosis in neurological applications.