The influence of a relaxation parameter on SPECT iterative reconstruction algorithms

Area of Science:

• Medical Imaging
• Computational Science

Background:

• Algebraic Reconstruction Techniques (ARTs) are crucial for image reconstruction.
• Image noise is a common issue in emission tomography reconstructions using ARTs.

Purpose of the Study:

• To investigate the impact of the ART relaxation parameter on image quality and convergence.
• To identify optimal relaxation parameter values for smoother emission tomography images.

Main Methods:

• Simulated and real single photon emission computed tomographic (SPECT) data were used.
• Scattering, attenuation, noise, and detector response were incorporated into simulations.
• Evaluated relaxation factors from 0.01 to 0.35 using metrics like CC, CV, CON, and SNR.

Main Results:

• Optimal relaxation factor and iteration count (around 0.1 and 8) balance key image quality metrics (CC, CV, CON, SNR).
• Exceeding optimal parameters increases contrast but degrades other metrics.
• Simulated results were validated with real SPECT phantom data.

Conclusions:

• The relaxation parameter critically influences ART image reconstruction quality in emission tomography.
• Specific parameter tuning is essential for achieving high-quality, low-noise reconstructed images.
• The simulation methodology accurately reflects real-world SPECT data performance.