This thesis is part of a large project concerning the development of a functional positron
emission tomography (PET) system for the study of the human brain, the so-called
BrainPET. Compared to the scanners that already exist on the market, this scanner should
provide a higher resolution and a higher sensitivity. This will allow more accurate
diagnosis of possible lesions, a lower radiological impact in the patient and better
functional and anatomical characterization of the brain when used in combination with a
magnetic resonance imaging (MRI) or computed tomography (CT) scanner. The
development of this scanner is a cooperation between the VUB (Belgium),
Forschungszentrum Jülich (Germany) and CIEMAT (Spain) under the Crystal Clear
The scanner will be constructed using a new concept of gamma radiation detectors, based
on monolithic detector blocks coupled to arrays of avalanche photodiodes. The outputs of
these arrays will be the inputs of a neural network (NN). After a training period, this NN
will provide the point of entrance of the gamma ray on the detector surface. The system
will be designed and tested to be compatible with intense magnetic fields allowing to
couple the scanner to a MRI system.
The work carried out included the development of the software that makes it possible to
implement a NN managing all the events on a computer. Moreover the possibility to use a
graphical card as a computing device for the NN algorithm, using CUDA as software
language, has been investigated.
An experimental setup was built to evaluate the energy resolution and spatial resolution for
different surface wrappings of the detector blocks. Also, the repeatability of the setup and
the energy response of the setup has been evaluated.
We managed to obtain the desired computing speed for the NN algorithm and made some
progress in the use of a graphical card as a computing device. A second experimental setup
was successfully built in order to allow the operation of two detector boxes in coincidence.
The energy response of different detector blocks measured with this setup was linear.
Moreover, it provided spatial and energy resolution results that were satisfying.
If you want to cite this thesis in your own thesis, paper, or report, use this format (APA):
CUYPERS, R. (2009). Data acquisition for monolithic scintillation detectors in PET applications.
Unpublished thesis, Xios, IWT.