Parameter Selection for Convex Optimization Time Calibration for a 2-Panel PET System.

Area of Science:

• Medical Imaging
• Nuclear Physics

Background:

• Coincidence timing calibration is essential for Positron Emission Tomography (PET) imaging.
• Factors like electronics, cable lengths, and detector physics introduce timing differences, increasing random rates and degrading performance.

Purpose of the Study:

• To investigate parameter selection for convex optimization using Ordinary Least Squares (OLS) for PET timing calibration.
• To evaluate the correlation between selected parameters and measured coincidence time differences.

Main Methods:

• Tested correlations between common parameters and experimental coincidence time differences.
• Evaluated 127 nested regression models to optimize Mean Squared Error (MSE), Bayesian Information Criterion (BIC), and Full Width at Half Maximum (FWHM).

Main Results:

• FWHM performance improved by approximately 53% across tested models.
• Timing resolution improved from ~304 ns to ~160 ns, still significantly deviating from the achievable ~10 ns FWHM for CZT detectors.

Conclusions:

• OLS optimization for timing calibration may be limited by the absence of critical parameters (e.g., trigger threshold, temperature) or excessive data variability.
• Further research is needed to identify optimal parameters or alternative calibration methods for improved PET timing resolution.