Machine learning is used to conduct positron emission particle tracking
Machine learning is used to conduct positron emission particle tracking lead image
Positron emission particle tracking (PEPT) is used to track the motion of a chosen particle, which has been tagged with a radioactive isotope, through 3D space. Over the years, several methods have been used to conduct PEPT, but each have their own limitations.
Andrei Leonard Nicusan and Christopher Robert Kit Windows-Yule used machine learning (ML) algorithms to conduct positron emission particle tracking.
Their method is highly scalable, can track multiple particles, has high temporal and spatial resolution, requires no prior knowledge of the number of tracers within the system, and can successfully distinguish multiple particles separated by distances as small as 2 millimeters.
“PEPT-ML can successfully locate a particle using fewer than 25 pairs of gamma rays. This facilitates the use of tracers with significantly lower activities than required for conventional PEPT, meaning that smaller tracers can potentially be used,” said Windows-Yule. “This opens up significant opportunities, for example, in the labelling and tracking of blood cells.”
PEPT-ML can be applied to any system that involves the flow of particles and/or fluids.
“We want to apply our method to study flow within ventricular assist devices, which can provide enhanced survival and quality of life for heart failure patients, but these devices currently carry risks such as pump thrombosis,” said Windows-Yule. “We hope that an improved understanding of flow within these devices may lead to improvements in their design and use, and hence improved results for patients.”
The authors are expanding this research by applying deep learning techniques to extract information regarding particles’ orientations from raw PEPT data. Their technique is freely available on the Birmingham Positron Imaging Centre’s repository.
Source: “Positron emission particle tracking using machine learning,” by A. L. Nicuşan and C. R. K. Windows-Yule, Review of Scientific Instruments (2020). The article can be accessed at https://doi.org/10.1063/1.5129251 .
