Virtual Noncontrast Images Derived From Photon-Counting CT Enhance Inter-Reader Reproducibility of Myocardial ECV Measurement in Patients With Cardiac Devices

OBJECTIVE: The standard 2-scan computed tomography (CT) for quantifying the extracellular volume fraction (ECV) of the left ventricle is limited by misregistration errors, particularly in patients with atrial fibrillation (AF) or cardiac implants such as prosthetic valves and/or cardiac implantable electronic devices (CIEDs). This study aimed to evaluate whether a single-scan virtual noncontrast (VNC) method using photon-counting CT (PCCT) improves the reproducibility of measurement compared with the standard subtraction method in these challenging patient populations. METHODS: We retrospectively analyzed 59 patients undergoing PCCT, stratified into the following 3 groups: control (n=20), AF (n=20), and cardiac implants, including prosthetic valves and/or CIEDs (n=19). Myocardial ECV was quantified using both standard subtraction and VNC methods. Reproducibility was assessed by intraclass correlation coefficients (ICCs) and mean absolute error. RESULTS: In the cardiac implant group, the VNC method demonstrated a significantly lower interobserver mean absolute error compared with the standard subtraction method (1.66% vs. 2.54%, P=0.029). No significant differences were observed in the AF (1.35% vs. 1.54%, P=0.648) or control (1.39% vs. 1.68%, P=0.674) groups. The VNC method demonstrated good to excellent interobserver (all ICCs >0.89) and intraobserver (all ICCs >0.90) agreement. Intraobserver agreement was also excellent for both methods, with no significant difference between them. CONCLUSIONS: The PCCT-derived VNC method significantly improves interobserver reproducibility of ECV measurement in patients with cardiac devices, establishing it as a more reproducible technique for myocardial tissue characterization in this challenging population.