Iterative reconstruction of PET images using a high-overrelaxation single-projection algorithm

Area of Science:

• Medical imaging
• Computational imaging
• Image processing

Background:

• Iterative image reconstruction is crucial for high-quality medical imaging.
• Optimizing reconstruction parameters can significantly improve efficiency and accuracy.
• Current methods may require extensive computation for large datasets.

Purpose of the Study:

• To develop a novel iterative image reconstruction procedure for enhanced precision.
• To accelerate the reconstruction process using optimized overrelaxation parameters.
• To demonstrate the generalizability of the derived parameters across various datasets.

Main Methods:

• Implemented an iterative image reconstruction algorithm.
• Utilized high initial overrelaxation parameters that decrease during iteration.
• Determined parameters pragmatically to maximize image quality gains.
• Validated the method with simulated and measured data.

Main Results:

• Achieved high-precision images within eight iterative steps.
• Demonstrated that parameters derived from one dataset are applicable to others.
• Showcased significant acceleration of iterative reconstruction.
• Confirmed the method's effectiveness for large datasets and 3D reconstruction.

Conclusions:

• The proposed iterative reconstruction method offers rapid, high-precision image calculation.
• Adaptive overrelaxation parameters effectively accelerate reconstruction and improve image quality.
• The parameter estimation strategy is robust and widely applicable, particularly for large-scale 3D imaging applications.