Magnetic resonance-guided positron emission tomography image reconstruction

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Biophysics

Background:

• Positron emission tomography (PET) image resolution is limited by physical processes and detector technology.
• Model-based reconstruction methods offer improvements over traditional filtered backprojection.
• Anatomical information integration is proposed to further enhance PET image quality.

Purpose of the Study:

• To review methods for incorporating anatomical information into PET image reconstruction.
• To assess the impact of anatomical information on PET image resolution, noise, and quantitation.
• To discuss the potential and limitations of MR-PET data for improving PET imaging.

Main Methods:

• Review of research over the past 20 years on anatomical information integration in PET reconstruction.
• Discussion of approaches using Markov random field priors.
• Exploration of methods based on joint information or entropy measures.

Main Results:

• Anatomical information integration can improve PET image resolution and noise properties.
• Combined MR-PET scanners provide intrinsically coregistered anatomical data.
• Methods discussed aim to enhance quantitation and lesion detection in PET imaging.

Conclusions:

• Incorporating anatomical information into PET reconstruction is feasible with combined scanners.
• Markov random field priors and joint information measures are key approaches.
• Further development of these methods holds significant potential for advancing PET imaging applications.