Improved Spectral Imaging with More Energy Bins for Realistic Photon Counting Detectors
Presentations
WE-G-201-4 (Wednesday, 7/13/2022) 2:45 PM - 3:45 PM [Eastern Time (GMT-4)]
Room 201
Purpose: Photon counting detectors hold great promise with their built-in spectral imaging capabilities. However, realistic detectors have imperfect energy responses that degrade their spectral imaging performance. Energy binning further reduces the amount of spectral information. We therefore investigate the importance of having more energy bins to maintain spectral imaging performance.
Methods: Two types of PCDs were investigated – silicon with 60 mm depth and CdTe with 1.6 mm depth. Their primary energy responses were obtained from previous Monte Carlo simulations and built into a prediction of their performance for various spectral imaging tasks (material decomposition (MD), virtual monoenergetic image (VMI)) of a 20 cm water, 1 cm calcium object as a function of the number of energy bins, from 2 to 8 bins. For each number of bins, the thresholds were optimized for each detector to minimize the average variance relative to an ideal PCD, as predicted by the Cramer-Rao lower bound. We also simulated CT acquisition of a head phantom with ideal, silicon, and CdTe PCDs to evaluate the impact of binning.
Results: For ideal, silicon, and CdTe PCDs, more energy bins are beneficial for spectral imaging tasks, especially MD. On average for all tasks, 8 bins reduces the variance by 55% for silicon and 19% for CdTe compared with 2 bins. The variance reduction is even larger for MD tasks, where silicon outperforms CdTe with 3 or more bins. CdTe has lower variance than silicon for VMI at 60 keV, although silicon with 8 bins is lower than CdTe with 2 bins. The predictions were confirmed in the simulated head phantom, which showed silicon with 8 bins had the lowest noise in the water MD image for realistic PCDs.
Conclusion: We demonstrated the importance of more energy bins for realistic PCDs, especially for a silicon PCD.
Funding Support, Disclosures, and Conflict of Interest: Funding from GE Healthcare
Keywords
Dual-energy Imaging, Noise, CT
Taxonomy
IM- CT: Dual Energy and Spectral
Contact Email