Photon-counting computed tomography of coronary and peripheral artery stents: a phantom study

Abstract

Accurate small vessel stent visualization using CT remains challenging. Photon-counting CT (PCD-CT) may help to overcome this issue. We systematically investigate PCD-CT impact on small vessel stent assessment compared to energy-integrating-CT (EID). 12 water-contrast agent filled stents (3.0-8 mm) were scanned with patient-equivalent phantom using clinical PCD-CT and EID-CT. Images were reconstructed using dedicated vascular kernels. Subjective image quality was evaluated by 5 radiologists independently (5-point Likert-scale; 5 = excellent). Objective image quality was evaluated by calculating multi-row intensity profiles including edge rise slope (ERS) and coefficient-of-variation (CV). Highest overall reading scores were found for PCD-CT-Bv56 (3.6[3.3-4.3]). In pairwise comparison, differences were significant for PCD-CT-Bv56 vs. EID-CT-Bv40 (p ≤ 0.04), for sharpness and blooming respectively (all p < 0.05). Highest diagnostic confidence was found for PCD-CT-Bv56 (p ≤ 0.2). ANOVA revealed a significant effect of kernel strength on ERS (p < 0.001). CV decreased with stronger PCD-CT kernels, reaching its lowest in PCD-CT-Bv56 and highest in EID-CT reconstruction (p ≤ 0.05). We are the first study to verify, by phantom setup adapted to real patient settings, PCD-CT with a sharp vascular kernel provides the most favorable image quality for small vessel stent imaging. PCD-CT may reduce the number of invasive coronary angiograms, however, more studies needed to apply our results in clinical practice.

Figure 1

Phantom setup with a water equivalent diameter of Dw = 27 cm and the stent positioned in the isocenter. The phantom was built to reflect the water-equivalent diameter of an average adult patient based on actual CT scans of a random sample of 457 individuals (189 women, 268 men; median age 61; range 19–93 years, median BMI 27.23; range 17.2–58.77). BMI, Body mass index.

Figure 2

Overview of the quantitative image analysis pipeline using an in-house developed MATLAB script. All stents were scanned on (a) PCD-CT and (b) EID-CT. (c) CT DICOM dataset of every stent was reconstructed, (d) 2D intensity profiles were calculated for each stent, (e) and plotted as 3D intensity plot across the entire length of the stent, (f) mean intensity profile of the 3D intensity profile, (g) automatic estimation of the ERS of every stent and kernel, (h) mean intensity-profile curve of the stent with the edge rise slope a HU of 10% and 90% of the maximum CT attenuation are shown. PCD-CT, Photon-counting detector CT; EID-CT, Energy integrating detector CT; DICOM, Digital imaging and communications in medicine; ERS, Edge rise slope; HU, Hounsfield Units

Figure 3

Result of the qualitative image analysis for (a) all stents and (b) stents limited to a diameter ≤ 3.5 mm. Results are presented as median and IQR. Results are shown for all readers. The p-values are calculated for the kernel comparison and presented as cross table for each image characteristic evaluation. All p-values are Bonferroni corrected for multiple comparisons. IQR, Interquartile ranges.

Figure 4

Coronary computed tomography angiography of a 62-year-old male patient with known coronary artery disease and stenting of the circumflex artery. Curved (a–c)) multiplanar reformations and axial (d–f)) reconstructed with different kernel strength (a and d) Bv40, (b and c) Bv48, (c and f) Bv56 depict the stent (2.5 mm diameter) in the circumflex artery. Stent lumen was best visible in the Br56 kernel reconstruction (c and f) and an in-stent restenosis could be reliably excluded. Scan was performed with 144 × 0.4 mm dual source, multi spectral high-pitch-flash-mode (3.2). Effective radiation dose was 1.07 mSv.

Figure 5

Image example of a 3 mm coronary artery stent (Everolimus-Eluting Platinum Chromium). Gray images depict the stent in EID-CT and three different kernel strengths of PCD-CT with the corresponding mean intensity profiles of the edge rise slope analysis. Subjective as well as objective analyses revealed higher image sharpness and reduced blooming with increasing kernel strength. PCD-CT, Photon-counting detector CT; EID-CT, Energy integrating detector CT.

Figure 6

Results of the quantitative image analysis. Panel (a) and (b) show results of the edge rise slope analysis. Results of the coefficient-of-variation analysis are presented in panel (c) and (d). ERS, Edge rise slope; CV, Coefficient-of variation.