Joint-MAP Bayesian tomographic reconstruction with a gamma-mixture prior.

Area of Science:

• Medical Imaging
• Computational Science
• Statistical Modeling

Background:

• Bayesian image reconstruction requires effective prior models to enhance image quality.
• Existing methods may struggle with capturing complex intensity distributions in tomographic data.
• A prior that models clustered intensity histograms can improve reconstruction accuracy.

Purpose of the Study:

• To develop a Bayesian image reconstruction method utilizing a mixture-of-gammas prior for clustered intensity histograms.
• To implement a joint-MAP (maximum a posteriori) estimation framework for improved tomographic image reconstruction.
• To demonstrate the efficacy of the proposed method in medical emission and transmission tomography.

Main Methods:

• Formulation of the problem within a joint-MAP estimation framework.
• Modeling the prior probability density function (PDF) as a mixture-of-gammas density.
• Iterative alternating descent optimization involving regularized likelihood estimation and histogram mixture decomposition.
• Transformation of objective functions for simplified optimization while preserving iteration fixed points.

Main Results:

• The mixture-of-gammas prior enforces positivity, a desirable property for image reconstruction.
• The iterative alternating descent effectively optimizes the complex objective functions.
• The method successfully handles the estimation of prior hyperparameters through mixture decomposition.
• Demonstrated applicability in both medical emission and transmission tomography.

Conclusions:

• The proposed Bayesian image reconstruction method effectively utilizes a clustered intensity histogram prior.
• The mixture-of-gammas density provides a robust and advantageous prior for tomographic imaging.
• The developed optimization strategy simplifies complex computations, making the method practical.
• The technique shows significant potential for advancing medical imaging reconstruction.