Timing Reconstruction in PMT-Based PET Detectors

Overview

Positron Emission Tomography (PET) is an important imaging technique in biomedical research and clinical diagnostics. Its performance is closely related to timing accuracy, which affects image quality and signal localization. This project focuses on waveform-level signal analysis in photomultiplier tube (PMT)-based detector systems, with the goal of improving timing precision through modeling and signal processing.

Rather than modifying hardware, the study explores how digital signal processing and statistical modeling can enhance timing estimation under realistic noise conditions. The work lies at the intersection of experimental physics, signal processing, and data analysis.

Detector and Signal Characteristics

The detector system consists of scintillation materials coupled with photomultiplier tubes (PMTs). When radiation interacts with the scintillator, optical photons are generated and collected by the PMT, producing electrical pulses. The temporal shape of these pulses contains information about photon arrival statistics, electronic response, and noise.

Methodology

The timing resolution depends on both signal slope and noise level. A simplified relationship can be expressed as:

\( \sigma_{t} \approx \frac{\sigma_{noise}}{\left| dV/dt \right|} \)

Based on this principle, the project models detector waveforms to separate different contributing factors, including photon arrival fluctuations and electronic noise. This provides a framework for evaluating timing extraction methods under controlled conditions.

Several signal processing techniques are implemented and compared:

• Constant Fraction Discrimination (CFD): Reduces amplitude-dependent timing variation.
• Leading-Edge Detection: Provides a simple baseline for timing extraction.
• Waveform Fitting: Uses parametric models to estimate the true signal onset more accurately.

My Contributions

• Data Analysis: Processed experimental waveform data to study signal variability under realistic conditions.
• Modeling: Built simplified computational models to reproduce signal formation and noise behavior.
• Algorithm Development: Implemented and compared multiple timing extraction methods.
• Evaluation: Analyzed how different processing strategies influence timing precision and system performance.

Results and Discussion

The results show that methods incorporating waveform shape information provide improved robustness compared to simple threshold-based approaches, especially in low signal-to-noise scenarios. This highlights the importance of combining physical modeling with signal processing techniques in detector optimization.

Outlook

Future work will focus on improving computational efficiency and exploring real-time implementation possibilities. The general approach developed in this project can also be extended to other sensing systems that rely on precise timing extraction from noisy signals.