Project Summary/Abstract
Positron emission tomography (PET) is a method of medical imaging that employs positron emitting
radionuclides attached to probe molecules (tracers) for non-invasively interrogating biological processes in vivo.
Each radionuclide decay emits a positron, which then combines with an electron and creates two oppositely
directed, colinear 511 keV annihilation photons. These annihilation photons are detected in opposing elements
of a photon detector ring forming lines of response (LOR) along which each positron emission originated. After
collecting millions of such photon pair events and positioning them along system LORs, an image can be
reconstructed to visualize and quantify the 3D distribution of tracer probe molecules within the body. Up to now,
PET systems detect only one tracer per study. However, more complete understanding of the disease biology
often requires the study of multiple biological processes simultaneously. Alzheimer’s Disease specifically is
characterized by presence of neuroinflammation, ß-amyloid, phosphor-t, and neurodegeneration. This project
aims to enable simultaneous imaging of multiple tracers by strategically choosing at least one PET tracer that
emits gamma photons in cascade with their positron. These gamma photons can be differentiated from
annihilation photons through their higher energy measured by the photon detector. Thus, LORs can be
associated with this positron + gamma tracer when a high energy photon arrives nearly the same time as a pair
of 511 keV photons.
Challenges associated with using prompt gamma emitters come from the lower probability of detecting both the
annihilation photons and gamma photon within the appropriate timing and energy windows. Another challenge
associated with using multiple tracers is misclassification of events among tracers due to missed, tissue
scattered, or random photons detected. Low detection efficiency and misclassification will reduce image quality
and accuracy of the associated multi-tracer images; thus, methods must be developed to mitigate these issues.
This project proposes to develop and characterize a position-sensitive endcap detector that will cover the open
end of an existing PET ring system to increase the detection efficiency for 3-photon events through increasing
the solid angle coverage of the system photon detectors. Signal processing algorithms will also be employed
using multiple temporal and energy windows to mitigate misclassification of photons coming from the two
emitters in addition to compensating for sensitivity differences between two- and three-photon emitters. These
techniques will include use of delayed time windows to estimate the different random coincidence rates and joint
maximum likelihood estimation of coincidence events based on system geometry. We will also use deep learning
to improve the image quality of the low dose images and to accurately separate the images obtained. Through
these techniques, this project aims to develop the first system capable of simultaneous multi-tracer PET imaging.