Spectral computed tomography (sCT) brings a promise of improved tissue discrimination when compared to conventional CT. At the heart of this new technology are energy selective photon counting detectors (PCD) combined with theories on how to select optimal energy bins for discriminating two or more materials. Several theories have been published on how to select these energy bins, but so far the diagnostic utility of optimised sCT has not been fully exploited.
This work presents a first step towards a virtual sCT scanner based on the well benchmarked BEAMnrc Monte Carlo code and the computer power of the University of Canterbury BlueFern supercomputer. A computational model of a recently developed sCT scanner (MARS-CT) has been developed to produce virtual X-ray projection data through an imaging object. The energy and position of all transmitted photons impinging on the detector plane can be extracted without the additional complications introduced by non-ideal behaviour (such as charge-sharing) of current detectors. The photons are grouped into selective energy bins to produce energy selective projection images of the imaging object (see figure 1). This enables the comparison of conventional CT with optimised spectral CT. Furthermore, the virtual sCT scanner is an ideal tool to compare and evaluate the different theoretical models (which optimise different metrics) in terms of relevant clinical parameters such as image contrast. In further work we are planning to include the physical limitations of the detector so the virtual sCT scanner closely resembles the MARS CT scanner.