Purpose: Multi-energy data from PCD-CT suffer from considerable spectral overlap. To improve energy separation and multi-energy CT performance, we propose a dual-source (DS)-photon-counting-detector (PCD)-CT approach, and determine the spectral settings and dose partition that provide optimal iodine quantification performance for different patient sizes.
Methods: Water phantoms of three sizes (25/35/45cm) containing iodine inserts (2.5/4/7.5/15mg/mL) were scanned to represent small/medium/large patients. A DS-PCD-CT was emulated by scanning twice using two tube potentials (kV) on a single-source whole-body PCD-CT with 2 energy thresholds. To determine the optimal kV-pair and energy thresholds, acquisition was performed with two kV-pairs (tube A/B=80/Sn140kV and 100/Sn140kV; Sn140 denotes 140kV with a Sn-filter) and different threshold combinations. The threshold-low for each PCD was fixed (25keV), with threshold-high varied from 50 keV to maximal value allowed (60/80/90keV for 80/100/140kV). Radiation dose was evenly split between the two acquisitions. Images were reconstructed and used for iodine quantification. The iodine root-mean-square-error (RMSE) was calculated to determine the optimal kV-pair and thresholds. Next, acquisition was repeated with different dose partitions between two tubes using the identified optimal kV and thresholds. Iodine RMSEs were calculated to optimize dose partition.
Results: Phantom results showed that a kV-pair of 80/Sn140kV, with energy thresholds around 25/55keV and 25/75keV, respectively, yielded optimal iodine RMSE for all phantom sizes. No significant artifact was observed in images acquired using this setup, even for the 45cm phantom. A dose ratio of 0.5:0.5 yielded the optimal or close-to optimal iodine RMSE for all three phantom sizes. The iodine RMSEs using the optimized setup were 1.07, 1.75, 2.80 mg/mL for three phantom sizes.
Conclusion: A DS-PCD approach was proposed to improve iodine quantification. Spectral acquisition setting and dose partition that provide optimal performance was identified based on phantom experiments. A universal setting yielded optimal or close-to optimal performance for different phantom sizes.