PET transmission tomography using a novel nonlocal MRF prior

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Computational Imaging

Background:

• Accurate attenuation correction is crucial for quantitative positron emission tomography (PET) reconstructions.
• Transmission scans estimate attenuation correction factors (ACFs), but short scan times introduce noise.
• Existing methods struggle with noise due to patient movement and discomfort from long scans.

Purpose of the Study:

• To apply a nonlocal prior Bayesian reconstruction method to PET transmission tomography.
• To address the challenge of noise in transmission scans caused by short acquisition times.
• To improve the quantitative accuracy of PET reconstructions by enhancing transmission image quality.

Main Methods:

• Utilized a nonlocal prior Bayesian reconstruction approach for PET transmission tomography.
• Assumed attenuation values vary smoothly with discontinuities at anatomical borders.
• Evaluated the method's performance under conditions of limited scan time and inherent noise.

Main Results:

• The nonlocal prior method demonstrated superior performance in reconstructing transmission images.
• The method effectively overcame noise effects, even with relatively short scan times.
• Reconstructed images showed improved quality and reduced noise compared to conventional methods.

Conclusions:

• Nonlocal prior Bayesian reconstruction is a robust method for PET transmission tomography.
• This approach enhances image quality and quantitative accuracy in the presence of noise.
• It offers a viable solution for reducing scan times while maintaining high-fidelity PET imaging.