Positron emission particle tracking using machine learning.

Area of Science:

• Physics
• Computer Science
• Nuclear Medicine

Background:

• Positron emission particle tracking is crucial for various scientific applications.
• Existing methods often struggle with simultaneous tracking of multiple particles or require prior knowledge of tracer numbers.

Purpose of the Study:

• To develop and demonstrate a new machine learning-based approach for positron emission particle tracking.
• To achieve high temporal and spatial resolution in particle localization and trajectory separation.
• To enable tracking of multiple particles without prior knowledge of their quantity.

Main Methods:

• Utilizing advanced machine learning algorithms for data analysis.
• Developing novel methods for particle location, tracking, and trajectory separation in 3D space.
• Implementing positron emission principles for tracer detection.

Main Results:

• Successfully located radioactively labeled particles in 3D space with high temporal and spatial resolution.
• Demonstrated the ability to distinguish multiple particles separated by as little as 2 mm.
• Showcased invariant spatial resolution regardless of the number of tracers used.

Conclusions:

• The developed machine learning approach offers a significant advancement in positron emission particle tracking.
• The technique allows for simultaneous tracking of a large number of particles without compromising data quality.
• This method opens new possibilities for complex systems requiring high-resolution particle tracking.