Pixel-based reconstruction (PBR) promising simultaneous techniques for CT reconstructions

Area of Science:

• Medical Imaging
• Computational Imaging
• X-ray Tomography

Background:

• Pixel-based reconstruction (PBR) algorithms are analogous to simultaneous iterative reconstruction techniques (SIRTs) used in X-ray transmission tomography.
• Previous studies indicated potential divergence issues with iterative algebraic algorithms in underdetermined systems.

Purpose of the Study:

• To discuss pixel-based reconstruction (PBR) algorithms for fan beam projections in X-ray transmission tomography.
• To investigate the stability and convergence of iterative algebraic algorithms using a refined pseudoprojection technique.

Main Methods:

• Simulation studies were performed using fan beam projections.
• The study evaluated iterative algebraic algorithms and convolution backprojection methods.
• Underdetermined conditions were simulated where object pixels outnumbered detector readings.

Main Results:

• Iterative algebraic algorithms demonstrated stability and did not diverge with the improved pseudoprojection technique.
• Convergence to the exact solution was observed and analyzed with varying numbers of projections.
• Reconstructions from convolution backprojection were provided for comparative analysis.

Conclusions:

• The refined pseudoprojection technique enhances the reliability of pixel-based reconstruction (PBR) algorithms in X-ray tomography.
• Iterative algebraic methods are viable for highly underdetermined systems in fan beam tomography.
• The number of projections significantly impacts reconstruction quality and convergence.