Maximum-Likelihood Estimation of Scintillation Pulse Timing.

Area of Science:

• Medical Imaging
• Nuclear Physics
• Signal Processing

Background:

• Positron Emission Tomography (PET) reconstruction benefits from accurate time-of-flight (TOF) information, enhancing signal-to-noise ratio.
• Waveform analysis is crucial for estimating timing accuracy, influenced by sampling frequency and data volume.
• Efficient data acquisition requires minimizing samples while retaining essential timing information for desired resolution.

Purpose of the Study:

• To develop an efficient data acquisition system for PET by optimizing waveform analysis.
• To introduce a maximum-likelihood (ML) estimation algorithm for precise digital pulse timestamping.
• To determine the minimum data required for achieving specific timing resolutions using Fisher Information (FI) analysis.

Main Methods:

• Developed a maximum-likelihood (ML) estimation algorithm using a contracting-grid search for timestamping digital pulses.
• Employed Fisher-Information (FI) matrix analysis to quantify timing information in waveform segments.
• Simulated waveforms for ML estimation and FI analysis, and validated timing resolution with experimental data from a LaBr3:Ce crystal and photomultiplier tubes.

Main Results:

• The ML estimation algorithm, implementable on FPGAs and GPUs, accurately assigns timestamps to digital pulses.
• FI analysis identified the minimal data segments necessary for achieving desired timing resolutions.
• Experimental data demonstrated that timing resolution plateaus as waveform segments lengthen, irrespective of sampling frequency beyond a certain point.

Conclusions:

• The developed ML algorithm and FI analysis provide a method for efficient data acquisition in TOF-PET.
• Optimized data acquisition reduces system complexity and cost without compromising timing performance.
• This approach enables the design of cost-effective and efficient PET systems with preserved timing resolution.