Simulated one-pass list-mode: an approach to on-the-fly system matrix calculation

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Computational Science

Background:

• Positron emission tomography (PET) system development requires robust image reconstruction algorithms adaptable to novel detector geometries and granular detection domains.
• Existing methods struggle with the dynamic and complex nature of prototype PET systems, necessitating on-the-fly calculations for measurement precision.

Purpose of the Study:

• To introduce a novel on-the-fly system matrix calculation method for list-mode PET reconstruction.
• To enhance flexibility in system modeling for complicated geometries and computational resources in prototype PET development.

Main Methods:

• Proposed a simulation of a one-pass list (SOPL) algorithm utilizing detection uncertainty models as random number generators.
• Employed ensembles of photon trajectories generated at image reconstruction time for each measurement datum.
• Investigated SOPL's behavior within the maximum likelihood-expectation maximization algorithm, altering the system matrix non-repetitively each iteration.

Main Results:

• SOPL demonstrated significantly enhanced noise properties in a two-dimensional imaging model compared to a non-random counterpart, despite a slight resolution penalty.
• The algorithm's noise reduction effectiveness can be improved by increasing ensemble samples, recovering resolution while retaining noise benefits.
• Validated SOPL against a standard system matrix using experimental data from the AX-PET prototype system.

Conclusions:

• The proposed SOPL algorithm offers a flexible and effective solution for image reconstruction in prototype PET systems with complex geometries.
• SOPL provides superior noise properties and adaptability, making it a valuable tool for advancing PET technology development.