Artifact Reduction From Dental Material in Photon-Counting Detector Computed Tomography Data Sets Based on High-keV Monoenergetic Imaging and Iterative Metal Artifact Reduction Reconstructions—Can We Combine the Best of Two Worlds?

Abstract

ObjectiveThe aim of this study was to compare the effectiveness of common strategies for artifact reduction of dental material in photon-counting detector computed tomography data sets.Materials and MethodsPatients with dental material who underwent clinically indicated CT of the neck were enrolled. Image series were reconstructed using a standard and sharp kernel, with and without iterative metal artifact reduction (IMAR) (Qr40, Qr40IMAR, Qr60, Qr60IMAR) at different virtual monoenergetic imaging (VMI) levels (40–190 keV). On representative slice positions with and without dental artifacts, mean and standard deviation of CT values were measured in all series at identical locations. The mean absolute error of CT values ( ∆falseHU¯) and the artifact index (AIX) were calculated and analyzed focusing on 3 main comparisons: (a) different VMI levels versus 70 keV, (b) standard versus sharp kernel, and (c) nonuse or use of IMAR reconstruction. The Wilcoxon test was used to assess differences for nonparametric data.ResultsThe final cohort comprised 50 patients. Artifact measures decreased for VMI levels >70 keV, yet only significantly so for reconstructions using IMAR (maximum reduction, 25%). The higher image noise of the sharp versus standard kernel is reflected in higher AIX values and is more pronounced in IMAR series (maximum increase, 38%). The most profound artifact reduction was observed for IMAR reconstructions (maximum reduction ∆falseHU¯: 84%; AIX: 90%).ConclusionsMetal artifacts caused by large amounts of dental material can be substantially reduced by IMAR, regardless of kernel choice or VMI settings. Increasing the keV level of VMI series, on the other hand, only slightly reduces dental artifacts; this effect, however, is additive to the benefit conferred by IMAR reconstructions.

Metal Artifact Reduction in Photon-Counting Detector CT

“…Generally, the results of this study are in agreement with clinical studies using PCD-CT: studies have shown that both iMAR and VMI can be effective in reducing metal artifacts from posterior spinal fixation, 19 dental implants, 17,18 or hip replacements. 20,21 These clinical studies also found best results for VMIs at an energy level of 110 keV.…”

“…[18][19][20] However, some studies did not see a significant reduction in artifacts with VMIs alone, which might be connected to the deviations in CT numbers also observed in this study. 17,18 Nonetheless, these studies still found additional benefits when combining iMAR and VMIs. In contrast to the results of Risch et al, 17 the current study did not find a significant impact of the reconstruction kernel, as the change in image noise was corrected for by the quantification algorithm.…”

“…17,18 Nonetheless, these studies still found additional benefits when combining iMAR and VMIs. In contrast to the results of Risch et al, 17 the current study did not find a significant impact of the reconstruction kernel, as the change in image noise was corrected for by the quantification algorithm.…”

“…The main limitation of the current study is that the results from the photon-counting CT scanner were not compared with a conventional CT scanner with energy-integrating detector. Although clinical data were not evaluated as part of this study, and clinical implants can be more complex in shape, size, and composition than the evaluated metal rods, the results show good agreement with previous clinical studies on metal artifact reduction in PCD-CT. [17][18][19][20][21] In addition, the evaluation was focused on ROI measurements in preselected positions. Further studies should also investigate the impact of metal artifacts on clinical tasks, for example, diagnostic performance.…”

“…Previous studies have found good performance for both iterative metal artifact reduction (iMAR) and VMI in clinical patient data. [17][18][19][20][21] However, patient studies have their own limitations as they do not allow for an investigation of different acquisition parameters (eg, tube potential) or different types of metal and lack a ground truth dataset for comparison. Furthermore, intraindividual differences between patients prohibit an accurate assessment of the impact of metal artifacts on quantitative CT-number accuracy in different tissues.…”