Characterization of a PET prototype based on monolithic detectors

We have characterized a PET prototype based on monolithic scintillator detectors. The prototype has a modular design with 8 identical detector cassettes, each housing 4 monolithic blocks. The cassettes are mounted on a gantry in 2 groups of 4 cassettes facing each other, totaling 32 detectors. Each detector block is composed of a trapezoidal LYSO:Ce scintillator glued to a pair of Hamamatsu S8550-02 APD matrices, which are read out by dedicated VATA 241 ASICs from Gamma Medica Ideas. The ASIC collects and amplifies the charge pulses from the 64 photosensors in each detector block, sums them into 8 rows and 8 column outputs and generates a trigger signal whenever the charge in a channel is above a user-defined threshold. Positioning algorithms based on supervised training neural networks are used for the detector blocks. A rotating platform with precise movement control is aligned with the system axis allowing acquisition of coincidence data along different projection angles. Using this system, Na22 point sources and a NEMA NU4-2008 image quality phantom filled with F18-FDG have been imaged and the images obtained have been characterized and compared with those from a Sedecal Argus PET/CT small-animal scanner, used as reference. Results show that the data acquisition system performs according to specifications and that PET imaging based on a large number of monolithic detector blocks is an alternative to designs based on segmented detectors.