Multiphase aortic valve calcium scoring on true-non-contrast and calcium-preserving spectral reconstructions using dual-source photon-counting detector CT.

Abstract

This study investigated differences between aortic valve calcium (AVC) scores derived from true non-contrast (TNC) and virtual-non-contrast reconstructions acquired with photon-counting detector CT (PCD-CT) and the impact of ECG-phase variability on AVC scores. A hundred patients undergoing PCD-CT for transcatheter aortic valve implantation (TAVI) planning were retrospectively analyzed. Scores were computed using the Agatston methodology for TNC and virtual-non-iodine (VNI) reconstruction at scanner-selected optimal phase (best) and a fixed ECG-phase (300 ms). For VNI reconstructions, additional phases from 150 ms to 450 ms with 50 ms increments were reconstructed. AVC scores of TNC_best vs TNC₃₀₀, VNI_best vs TNC_best, VNI₃₀₀ vs TNC₃₀₀, and all VNI phases vs VNI_best were compared using Wilcoxon signed-rank tests. The agreement was assessed using scatter plots, Bland-Altman plots, and intra-class coefficients. AVC scores were also categorized based on the likelihood of severe aortic stenosis. Differences between reconstructions were evaluated as percentages (reclassification) and analyzed using Cohen's kappa coefficients. TNC_best and TNC₃₀₀ differed significantly (mean bias: 226; LoA: [-820, 1300]; p < 0.001, reclassification 17%). VNI_best vs TNC_best resulted in a mean bias of -512 (LoA: [-1900, 860]; p < 0.001) and reclassification of 17%. TNC₃₀₀ vs VNI₃₀₀ demonstrated a bias of -200 and reclassification of 14% (κ = 0.72). VNI reconstructions showed less variability across phases than the difference between TNC_best and TNC₃₀₀ (range, mean bias: 22-146). VNI is a feasible alternative for AVC scoring but tends to overestimate compared to TNC. While phase-dependent variability in TNC underscores the need for standardization, further optimization of VNI is necessary for routine clinical use. Question What is the performance of AVC score calculation from virtual non-contrast images with PCD-CT and the impact of the reconstructed ECG-phase? Findings VNI reconstructions tend to overestimate the scores compared to non-enhanced acquisitions. ECG phase significantly impacts AVC scores for non-enhanced acquisitions and VNI reconstructions. Clinical relevance Utilizing VNI reconstructions to calculate AVC scores might reduce radiation dose, and understanding the influence of ECG-phase on these scores might improve reliability.