Evaluation of diffusion models in breast cancer

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Biomedical Engineering
• Radiology

Background:

• Diffusion-weighted MRI (DW-MRI) is crucial for characterizing tissue microstructure.
• Non-monoexponential diffusion behavior may indicate microvascular pseudodiffusion and tumor heterogeneity.
• Understanding these effects is vital for accurate breast cancer assessment.

Purpose of the Study:

• To investigate microvascular pseudodiffusion and non-monoexponential behavior in breast cancer.
• To assess the impact of tumor spatial heterogeneity on diffusion MRI signals.
• To explore methodological factors influencing diffusion MRI in phantom studies.

Main Methods:

• Diffusion MRI sequences with multiple b-values were used.
• Eddy current effects and b-value accuracy were evaluated in phantoms.
• Diffusion models (monoexponential, stretched-exponential, biexponential IVIM) were fitted to data from volunteers and patients using the Bayesian information criterion.

Main Results:

• 54.3% of lesion voxels fitted a monoexponential model; 26.2% a stretched-exponential; 19.5% a biexponential IVIM model.
• High correlation between monoexponential and stretched-exponential diffusion coefficients was found (0.92).
• Eddy current distortions varied with diffusion gradient direction, impacting high b-value images.

Conclusions:

• Microvascular pseudodiffusion and intravoxel heterogeneity were not consistently observed in breast cancer voxels.
• Stretched-exponential and IVIM models are more suitable for regional analysis than global tumor assessment.
• Geometric averaging of diffusion-weighted images acquired with opposite diffusion gradients is recommended for clinical protocols.