PET with a Dual-Head Coincidence Camera: Spatial Resolution, Scatter Fraction, and Sensitivity

Scintillation cameras with options for detecting positron annihilation quanta in the coincidence acquisition mode may be the most innovative diagnostic devices introduced in nuclear medicine during the last few years. Besides conventional low-energy imaging in the collimated single-photon mode, these options offer a relatively inexpensive opportunity to perform uncollimated PET by switching into the coincidence acquisition mode. Instead of collimators, scatter frames (with 2 optional configurations: axial or open scatter frame) can be mounted to reduce the amount of quanta reaching the detectors from parts of the patient's body outside the field of view. This study investigates the coincidence imaging properties of the scintillation camera by measuring spatial resolution, scatter fraction, sensitivity, and count-rate response for 18F. A needle in air and a plastic tube in water, each filled with 18F, were oriented axially and transversally to measure the transverse and axial spatial resolutions, respectively. Using either the axial or the open scatter frame, a standard cylinder filled homogeneously with activity was studied over several half-life periods to deduce the respective scatter and random fractions of the system by means of a sinogram technique. The activity of the cylinder was kept low to determine the sensitivity to coincidence events for both scatter frames. Depending on the distance between the line source and the axis of rotation and on the choice of the axial acceptance angle used to reconstruct the coincidence events (single-slice rebinning algorithm), the axial resolution was found to be between 6 and 10 mm (full width at half maximum) with the axial scatter frame mounted. The transversal resolution was 6-6.5 mm on the axis of rotation, independent of the scatter frame used. The scatter fraction amounted to roughly 25% for the axial and 38% for the open scatter frames. The sensitivity when measuring true coincidence pairs ran to nearly 650 Hz/kBq/mL, when acquisition was performed with the axial scatter frame using a 30%-wide photopeak energy window. When acquiring with the open scatter frame, the sensitivity increased to nearly 3000 Hz/kBq/mL. Using the axial scatter frame, the homogeneously filled cylinder could be scanned with a maximum true coincidence rate of 2000 Hz for an activity of 55-60 MBq. Although this maximum true coincidence counting rate did not change significantly when the acquisition was performed with the open scatter f