On the Optimal Number of Energy Channels in Multi-Energy Photon-Counting Based CT to Improve Gold Identification: A Simulation Study
X Hu1, Y Zhong1, K Yang2*, X Jia1, (1) University of Texas Southwestern Medical Center, Dallas, TX, (2) Harvard Medical School, Massachusetts General Hospital, Boston, MA
Presentations
SU-I400-BReP-F1-1 (Sunday, 7/10/2022) 4:00 PM - 5:00 PM [Eastern Time (GMT-4)]
Exhibit Hall | Forum 1
Purpose: Material decomposition is one of the major applications of photon-counting detector (PCD)-based multi-energy CT (MECT). PCD often operates in a mode acquiring photon counts above a user-defined energy threshold. For PCD-CT, it requires switching this threshold among projections to acquire data with multiple energy channels. As the number of channel increases, the increased information along energy dimension favors material decomposition, whereas the undersampling along projection dimension increases challenges in image reconstruction. We hypothesize that there is an optimal number of energy channels Nₑ for material decomposition. This study numerically investigates the optimal Nₑ for decomposition with gold contrast.
Methods: We considered PCD-based MECT data acquisition under a 100 kV x-ray beam for a CT insert phantom containing inserts with solutions with 0.1-1.0 % gold contrast agent. We fixed the total number of projections to 1800 in a single rotation. For a given Nₑ, we selected energy thresholds by balancing photons counts among all channels. We simulated data acquisition with energy threshold sequentially sweeping among projections. We developed a model to simultaneously reconstruct CT images and decompose them into three base materials (water, bone, and gold). Alternative Direction Method of Multipliers (ADMM) was employed to solve the optimization problem. We computed decomposition errors, properties of projection matrices and decomposition matrices, as a function of Nₑ.
Results: Condition number of projection matrix increases and that of decomposition matrix decreases with Nₑ. Gold decomposition accuracy of relatively high concentration (1%) was not affected by Nₑ. However, the best decomposition for low-concentration cases occurred at Nₑ=5-7.
Conclusion: There is likely an optimal number of energy channels in PCD-based MECT for the task of gold material decomposition at low concentration due to trade-off between numerical properties of projection matrix and decomposition matrix.
Funding Support, Disclosures, and Conflict of Interest: This study was supported in part by grants from National Institutes of Health (R37CA214639, R01CA227289).
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email