Positron emission tomography (PET) is a medical imaging technique that produces a three-­‐dimensional image
of functionalprocesses inthe body.
The system detects pairs of gamma rays emitted indirectly by a positron-­‐emitting radionuclide,
which is introduced into the body on a biologically active molecule. Multimodality imaging and,
more specifically, the combination of PET and CT has matured into an important diagnostic
tool. During the same period, concepts for PET scanners integrated into a magnetic resonance
(MR) tomography have emerged. The purpose of this project is to develop the framework that will
allow the use of monolithic crystal to build time-of-flight detectors for whole body PET.
The main limitations for using monolithic crystal are produced by the complexity of the
calibration methods available nowadays and the poor spatial resolution obtained far for the
photosensor.
In this project we aim to simplify the calibration procedure by using unsupervised learning
methods based on training data. That method will allow not only to simplify the calibration
procedure at the factory but also to make a follow-up of the calibration status.
In order to improve the spatial resolution different approached will be studied.
In particular, we propose the use of phosphor reflector to decouple forward and reflected
light. In that way spatial resolution can be improved while other performance parameter are
not degraded. In addition, special attention will be paid on making it compatible for combined
PET and MR imaging by using novel silicon photomultiplier (SiPM) devices. In particular,
SiPM completely scalable and with improved fill factor will be tested. New digital SiPM
will be tested on this project as well. Finally, further comparison on the performance of monolithic
and pixelated detectors will be performed under same detector configuration.