PET Image Reconstruction and Deformable Motion Correction Using Unorganized Point Clouds.

Area of Science:

• Medical Imaging
• Image Reconstruction
• Positron Emission Tomography (PET)

Background:

• Quantitative PET imaging necessitates motion correction for accurate results.
• Traditional methods struggle with non-cyclic deformable motion due to the high number of gates required.
• Existing techniques may be impractical for complex, non-cyclic motion scenarios.

Purpose of the Study:

• To develop and validate an alternative iterative image reconstruction approach for PET.
• To address the limitations of gate-based motion correction in the presence of non-cyclic deformable motion.
• To introduce a method utilizing point clouds for explicit motion correction in PET.

Main Methods:

• Proposed an iterative image reconstruction method using unorganized point clouds to model objects.
• Incorporated time-dependence into point coordinates for explicit deformable motion correction.
• Utilized Voronoi cells for defining basis functions and validated with MLEM and OSEM algorithms.

Main Results:

• Demonstrated quantitative accuracy and stability in reconstructed images from digital and physical phantoms.
• Showcased comparable noise and convergence properties to traditional basis functions.
• Successfully validated point-cloud-based MLEM and OSEM algorithms for PET motion correction.

Conclusions:

• The proposed point-cloud-based approach offers a viable alternative for motion correction in quantitative PET.
• This method effectively handles non-cyclic deformable motion, overcoming limitations of traditional gating techniques.
• The approach provides quantitatively stable and accurate image reconstructions with favorable noise and convergence characteristics.