No abstract

A small diameter positron emission tomography, designed specifically for small animal studies, was constructed from existing, commercially available, bismuth germanate (BGO) detectors and electronics. The scanner consists of 16 BGO detector blocks arranged to give a tomograph with a diameter of 115 mm and an axial field of view (FOV) of 50 mm. Each block is cut to produce eight (axial) by seven (radial) individual detector elements. The absence of interplane septa enables the acquisition of 3D data sets consisting of 64 sinograms. A 2D data set of 15 sinograms, consisting of eight direct and seven adjacent cross planes, can be extracted from the 3D data set. Images are reconstructed from the 2D sinograms using a conventional filtered backprojection algorithm. Two methods of normalization were investigated, based on either a rotating 68Ge rod source, or a uniform 68Ge plane source, with a uniform cylindrical 18F phantom. Attenuation of the emitted photons was estimated using a rotating 68Ge rod source. The transaxial resolution of the tomograph was measured as 2.3 mm full width at half maximum (FWHM) and 5.6 mm full width at tenth maximum (FWTM) at the centre of the FOV, degrading to 6.6 mm (radial) and 4.4 mm (tangential) FWHM and 10.4 mm (radial) and 14.4 mm (tangential) FWTM at 40.0 mm from the centre of the FOV. The axial slice width was 4.3 mm FWHM, 10.3 mm FWTM at the centre of the transaxial field of view and 4.4 mm FWHM, 10.6 mm FWTM at 20.0 mm from the centre of the FOV. A scatter fraction of 31.0% was measured at 250-850 keV, for an 18F line source centred in a 60 mm diameter, water-filled phantom, reducing to 20.4% and 13.8% as the lower energy discrimination was increased to 380 keV and 450 keV, respectively. The count rate performance was measured using a noise equivalent count rate method, and the linearity of the dead time correction was confirmed over the count rates encountered during routine scanning. In 2D mode, the absolute sensitivity of the tomograph was measured as 9948 counts s-1 MBq-1 at 250-850 keV, 8284 counts s-1 MBq-1 at 380-850 keV and 6280 counts s-1 MBq-1 at 450-850 keV.

Abstract--We present construction methods and performance results for a production scintillator array of 6 4 optically isolated, 3 mm x 3 mm x 30 mm sized LSO crystals. This scintillator array has been developed for a PET detector module consisting of the 8x8 LSO array coupled on one end to a single photomultiplier tube (PMT) and on the opposite end to a 64 pixel array of silicon photodiodes (PD). The PMT provides an accurate timing pulse and initial energy discrimination, the PD identifies the crystal of interaction, the sum provides a total energy signal, and the PD/(PD+PMT) ratio determines the depth of interaction (DOI). Unlike the previous LSO array prototypes, we now glue Lumirror reflector material directly onto 4 sides of each crystal to obtain an easily manufactured, mechanically rugged array with our desired depth dependence. With 511 keV excitation, we obtain a total energy signal of 3600 electrons, pulse-height resolution of 25% fwhm, and 6-15 mm fwhm DOI resolution.

The measurement of depth of interaction (DOI) within detectors is necessary to improve resolution uniformity across the FOV of small diameter PET scanners. DO1 encoding by pulse shape discrimination (PSD) has definite advantages as it requires only one readout per pixel and it allows DO1 measurement of photoelectric and Compton events. The PSD time characteristics of various scintillators were studied with avalanche photodiodes (APD) and the identification capability was tested in multi-crystal assemblies with up to four scintillators. In the PSD time spectrum of an APD-GSO/LSO/ BGO/CsI(Tl) assembly, four distinct time peaks at 45, 26, 88 and 150 ns relative to a fast test pulse, having resolution of 10.6, 5.2, 20 and 27 ns, can be easily separated. Whereas the number and position of scintillators in the multi-crystal assemblies affect detector performance, the ability to identify crystals is not compromised. Compton events have a significant effect on PSD accuracy, suggesting that photopeak energy gating should be used for better crystal identification. However, more sophisticated PSD techniques using parametric time-energy histograms can also improve crystal identification in cases where PSD time or energy discrimination alone is inadequate. These results c o n f i i the feasibility of PSD DO1 encoding with APD-based detectors for PET.

No abstract