Strategies to increase the signal to noise ratio in three-dimensional positron emission tomography.
Positron Emission Tomography (PET) is an imaging technique that uses biologically
relevant molecules labelled with positron emitting radioisotopes to measure regional tissue
function in living organisms. To maximise the detection efficiency, data are acquired in 3D,
that is, all possible detector combinations in a scanner without inter-ring shielding (septa).
The gain in sensitivity afforded by 3D PET is offset by the increase in random coincidences,
scattered coincidences and deadtime. These problems must be overcome for the gain in
sensitivity to be fully realised. The aim of this research project was to investigate strategies to
increase the signal to noise ratio of the 3D PET data.
Additional side shielding, both in neuro and body scanning, has been implemented
and assessed. Large gains were achieved using the neuro shields in experimental and clinical
studies. The potential of the body shields was tested in experimental and in-vivo studies
which showed that they were scan dependent. For example, no gain was found for a cardiac
blood flow (H2
A model-based scatter correction was assessed by companng compartment ratios
within the 'Utah' phantom with radioactivity outside the field of view, with and without neuroshielding.
Recovered ratios were within 6% of their actual values.
The integration time was reduced in an effort to decrease the system deadtime. A peak
increase of 150/0 in noise equivalent count rate was measured for a uniform cylinder inside the
field of view.
A random coincidence variance reduction technique was implemented and assessed to
reduce the noise contained in the delayed window random coincidence estimate. The
algorithm was evaluated using phantoms and tested on clinical data. A mean 16% reduction in
coefficient of variation was measured for a C15O torso study.