Evaluation of a CZT gamma-ray detection module concept for SPECT

CZT detectors are the basic building block of a variety of new SPECT systems. Their modularity allows to adapt system architecture to specific applications such as cardiac, breast, brain or small animal imaging. Thanks to their high quantum yield, these direct conversion detectors exhibit better energy and spatial resolutions than usual scintillation detectors based on NaI(TI). However, it remains often unclear if SPECT imaging really can take profit of that performance gain. We propose here to conduct a case study based on the latest results obtained in our laboratory with current state of the art ICs and CZT crystals to investigate the system performance of a classical module dimensioning of 5 mm thick CZT with a segmented anode at a 2.5 × 2.5 mm pitch. This dimensioning, though being quite conservative, allows an easy integration in terms of crystal hybridization or PCB layout but still allows to obtain impressive results. Compared with X-ray counting were the only information retrieved is the occurrence of a photon interaction, spectrometric imaging performance is not only determined by photon statistics but also by readout noise, that ultimately limits time, energy and spatial resolutions associated with each photon event. After a first part dealing with the noise performance achieved by integrated circuits and a discussion on the key limiting factors, we present the typical readout architecture used and show how the signal processing is optimized for multiple parameter estimation. In a subsequent part, multi-pixel data acquisition scheme is discussed to show how raw channel data is used to extract photon parameters (energy, timestamp, and 3D position) while taking into account material non-uniformities. We will show how such data can be used to build images and quantify resulting improvement. Finally, we open the discussion on SPECT collimation and architecture by demonstrating with simulations how a tomographical system dimensioning depends on detect- r performance.