Room: AAPM ePoster Library
Purpose: Spectroscopic x-ray detectors (SXDs) are gaining attention for their ability to perform single-shot, multi-material decomposition. However, SXDs suffer from cross-talk between neighbouring detector elements and energy bins due to charge diffusion, Coulomb repulsion and characteristic emission. This spatio-energetic cross-talk, referred to as charge sharing, results in spectral distortion, which degrades image quality. The purpose of this work was to investigate the maximum likelihood expectation-maximization (MLEM) algorithm for correction of charge sharing in SXDs.
Methods: Charge sharing leads to high-energy photons being erroneously recorded in low-energy bins of detectors elements surrounding those in which primary interactions occur. We formulated this as maximum likelihood problem and used the MLEM algorithm to put the erroneous counts back in the correct energy bins and detector elements. We investigated this approach by simulation. To this end, flat-field x-ray images were simulated with 2 energy bins for a range of x-ray spectra with half-value layers ranging from 5.8 to 12.1 mm of aluminum. The simulations accounted for diffusion, Coulomb repulsion and K-shell characteristic emission/reabsorption. Our simulation model was calibrated against a CZT-based SXD. The correction algorithm was applied to the simulated images and its accuracy was assessed in terms of the percent error between the true number of counts and the MLEM-based estimates of the counts.
Results: The EM algorithm successfully corrected the number of photons with an accuracy of > 80% for all energy bins across all x-ray spectra included in this analysis.
Conclusion: The EM algorithm has the potential to correct for charge sharing in SXDs. A focus of future work will be to investigate the performance of the algorithm of systems that use 3 or more energy bins, and to verify algorithm performance experimentally.
Funding Support, Disclosures, and Conflict of Interest: This research was funded by the Natural Sciences and Engineering Council of Canada (NSERC).