Bayesian modeling with locally adaptive prior parameters in small animal imaging

Area of Science:

• Medical Imaging
• Statistical Modeling
• Image Processing

Background:

• Medical images often suffer from noise and low resolution, hindering accurate measurements.
• Image noise suppression and resolution enhancement are critical inverse problems in medical imaging.
• Existing methods may struggle with limited data and numerous unknowns.

Purpose of the Study:

• To develop novel, robust statistical estimation approaches for inverse problems in image processing and reconstruction.
• To implement and analyze a locally adaptive Markov chain Monte Carlo algorithm.
• To improve the accuracy and reliability of medical imaging analysis.

Main Methods:

• Implementation of Bayesian methods, specifically a locally adaptive Markov chain Monte Carlo (MCMC) algorithm.
• Analysis of algorithm robustness through parameter variation and diverse experimental setups.
• Application to radionuclide imaging using a prototype gamma camera with simulated data.

Main Results:

• The locally adaptive MCMC algorithm demonstrated superior edge recovery compared to non-locally adaptive methods.
• The proposed method significantly reduced estimation uncertainty and bias.
• Locally adaptive smoothing improved estimation accuracy over homogeneous Bayesian models.

Conclusions:

• The locally adaptive MCMC algorithm offers flexibility and robustness for medical imaging applications.
• Improved image quality leads to more reliable interpretation and quantification.
• This approach enhances the precision of measurements from medical imaging data.