Interpretation of diffusion MR imaging data using a gamma distribution model.

Area of Science:

• Biomedical Imaging
• Diffusion MRI Physics
• Quantitative Histopathology

Background:

• Interpreting non-Gaussian diffusion MRI data in biological tissues remains challenging for correlating with histological changes.
• Statistical models, assuming continuous distribution of diffusion coefficients within a voxel, offer better histological context than other models.
• The gamma distribution model is proposed for enhanced interpretation of diffusion MRI data.

Purpose of the Study:

• To evaluate a statistical diffusion MRI model based on the gamma distribution for improved histological interpretation.
• To compare the goodness of fit and diagnostic capability of the gamma model against other diffusion models.
• To assess the correlation between diffusion MRI parameters and histological features in prostate cancer.

Main Methods:

• Compared the gamma model, bi-exponential model, and truncated Gaussian model for goodness of fit.
• Evaluated area fractions of specific diffusion coefficient (D) ranges to assess diagnostic capability.
• Utilized a clinical dataset of histologically proven prostate cancer cases.

Main Results:

• The gamma model demonstrated a better fit for cancer tissue compared to the truncated Gaussian model; no significant difference was found between the gamma and bi-exponential models.
• No significant differences in model fit were observed for normal peripheral zone tissue across all models.
• A 2D scatter plot of area fraction for D < 1.0 mm²/s (frac < 1) versus D > 3.0 mm²/s (frac > 3) clearly separated cancer and non-cancer tissues.

Conclusions:

• The proposed gamma distribution model provides a good fit for diffusion MRI data.
• The model facilitates a more direct histological interpretation of diffusion MRI findings.
• This approach shows promise for differentiating cancerous from non-cancerous tissues based on diffusion characteristics.