Beam hardening of K-edge contrast agents: a phantom study comparing clinical energy-integrating detector and photon-counting detector CT systems

Abstract

Background: Beam hardening (BH) artifacts negatively influence computed tomography (CT) measurements, especially when due to dense materials or materials with high effective atomic numbers. Photon-counting detectors (PCD) are more susceptible to BH due to equal weighting of photons regardless of their energies. The problem is further confounded by the use of contrast agents (CAs) with K-edge in the diagnostic CT energy range. We quantified the BH effect of different materials comparing energy-integrating detector (EID)-CT and PCD-CT.

Methods: Pairs of test tubes were filled with dense CA (iodine-, gadolinium-, and bismuth-based) and placed inside a water phantom. The phantoms were scanned on EID- and PCD-CT systems, at all available tube voltages for the PCD scanner. Images were reconstructed with standard water BH correction but without any iodine/bone BH corrections. Virtual monoenergetic images (VMI) were calculated from PCD-CT data.

Results: PCD-CT had higher CT numbers in all x-ray spectra for all CAs (p < 0.001) and produced larger cupping artifacts in all test cases (p < 0.001). Bismuth-based CA artifacts were 3- to 5-fold smaller than those of iodine- or gadolinium-based CA. PCD-CT-based VMI completely removed iodine BH artifacts. Iodine BH artifacts decreased with increasing tube voltage. However, gadolinium-based BH artifacts had a different trend increasing at 120 kVp.

Conclusion: EID had fewer BH artifacts compared to PCD at x-ray tube voltages of 120 kVp and higher. The inherent spectral information of PCDs can be used to eliminate BH artifacts. Special care is needed to correct BH artifacts for gadolinium- and bismuth-based CAs.

Relevance statement: With the increasing availability of clinical photon-counting CT systems offering the possibility of dual contrast imaging capabilities, addressing and comprehending the BH artifacts attributed to old and novel CT CAs grows in research and ultimately clinical relevance.

Key points: EID-CT provides fewer BH artifacts compared to PCD-CT at x-ray tube voltages of 120 kVp and higher. K-edge CAs, such as those based on gadolinium, further confound BH artifacts. The inherent spectral information of photon counting detector CT can be used to effectively eliminate BH artifacts.

Keywords: Artifacts; Bismuth; Gadolinium; Iodine; Tomography (x-ray computed).

Fig. 1

Sample 120-kVp EID image of the IBCA test object with iodine/bone BH correction intentionally turned off to accentuate the artifacts. a BH cupping artifact is visible as a decrease in the CT number of dense columns of IBCA. b BH shadow artifact presents a dark streak between the two IBCA inserts. The line profile shows both cupping and shadow artifacts. c Schematic of cupping artifact which is measured as the difference in average CT numbers of a circular ROI (#1) (diameter 0.75 cm) placed at the center of the test tube and a ring ROI (#2) (1.5 < diameter < 2.25 cm). d Shadow artifact quantified as the difference between the CT number of an ROI of 0.75 cm in diameter placed in the center of the shadow artifact (#3) and the CT number of background water (#4). CT, Computed tomography; IBCA, Iodine-based contrast agent; ROI, Region of interest

Fig. 2

Simulated normalized detected x-ray spectra showing the relative contribution of photons of different energies to the signal detected by EIDs (solid red line) and PCD (dashed blue line) for a projection passing through 11 cm of water and 0.2-mm equivalent of ‘solid’ CAs. The effect of the K-edge of gadolinium and bismuth on the detected spectra is annotated with the black arrows

Fig. 3

a–c Mean CT numbers of CAs used in this study with respect to different tube voltage and prefiltration settings for EID and PCD images. PCD CT numbers were significantly higher than those of EID for all tube settings and all CAs (p < 0.001). d–f CNR between water and the CAs in (a–c) IBCA, GBCA, and BBCA. PCD showed significantly higher CNR for all tube settings and all CAs (p < 0.004). BBCA, Bismuth-based contrast agent; CNR, Contrast-to-noise ratio; CT, Computed tomography; EID, Energy-integrating detector; GBCA, Gadolinium-based contrast agent; IBCA, Iodine-based contrast agent; PCD, Photon-counting detector

Fig. 4

Cropped samples of all images used in this study showing the shadow BH artifacts for different CAs at different x-ray tube settings. Iodine/bone BH correction was intentionally turned off to accentuate the artifacts. BBCA, Bismuth-based contrast agent; EID, Energy-integrating detector; GBCA, Gadolinium-based contrast agent; IBCA, Iodine-based contrast agent; PCD, Photon-counting detector

Fig. 5

a–c Cupping BH artifact induced by the three CAs at different x-ray tube settings for EID and PCD images. Iodine/bone BH correction was intentionally turned off to accentuate the artifacts. The PCD images showed significantly higher artifacts across all CAs and tube settings (p < 0.004). d–f Shadow BH artifact induced by IBCA was significantly higher in PCD images for all tube settings except for 90 kVp, due to severe artifacts that saturated the CT number to -1,024 HU. Iodine/bone BH correction was intentionally turned off to accentuate the artifacts. Shadow artifacts induced by GBCA were significantly higher for PCD (p < 0.001). The magnitude of artifacts induced by BBCA was very small and not significantly different between the two detector systems, except at 100Sn (p < 0.008). BBCA, Bismuth-based contrast agent; CT, Computed tomography; EID, Energy-integrating detector; GBCA, Gadolinium-based contrast agent; IBCA, Iodine-based contrast agent; PCD, Photon-counting detector

Fig. 6

a Line profiles of PCD-derived VMIs and threshold-1 and threshold-2 images connecting the centers of two IBCA contrast columns showing cupping and shadow artifacts according to Fig. 1b. Iodine/bone BH correction was intentionally turned off to accentuate the artifacts. The cup shape flattens as the VMI energy increases. After a certain VMI energy, the flat shape is overcorrected into a cap. This is denoted by negative BH metrics in (b, c). Quantitative assessment of cupping (b) and shadow (c) artifacts with respect to VMI energy for IBCA and GBCA shows a decrease in the artifact with the increase of VMI energy, followed by overcorrection. CT, Computed tomography; EID, Energy-integrating detector; GBCA, Gadolinium-based contrast agent; IBCA, Iodine-based contrast agent; PCD, Photon-counting detector; VMI, Virtual monoenergetic image