Diagnostic Concordance of PCCT Virtual Non-contrast Imaging for Hepatic Steatosis Assessment: A Comparative Study of IQ155 and IQ200 Protocols

Objective: To evaluate the accuracy of photon-counting CT (PCCT) virtual non-contrast (VNC) imaging in quantifying hepatic steatosis using two radiation dose protocols (IQ155 vs. IQ200). Methods: 146 consecutive patients (March-May 2025) were prospectively enrolled (IQ155: n=86; IQ200: n=60). Hepatic fat content was classified into four PDFF-based grades: normal (<5%), mild (5-14%), moderate (15-24%), and severe (≥25%). VNC attenuation error (VNCerror) was calculated for arterial (AP) and venous (VP) phases. Liver-to-spleen (L/S) ratio concordance between true non-contrast (TNC) and VNC reconstructions was assessed via Bland-Altman analysis. Results: No significant inter-protocol differences in VNCerror between different dose (IQ155 vs IQ200) (AP: P =0.62; VP: P =0.36). Significant VNCerror differences between different CTFF groups (AP: P<0.05; VP: P<0.01) in IQ155. No significant inter-grade differences (AP: P =0.059; VP: P =0.147) despite similar trends in IQ200. Significant positive VNCerror-CTFF correlations in all phases except IQ200 VNCVp (IQ155-AP: r=0.22, P=0.0035; VP: r=0.32, P<0.001; IQ200-AP: r=0.31, P<0.001; VP: r=0.32, P=0.14). Regarding L/S Concordance, Mean bias AP -0.032 (95% LoA: -0.361–0.297), VP -0.013 (95% LoA: -0.328–0.301) in IQ155 group. Mean bias AP -0.092 (95% LoA: -0.357–0.173), VP -0.062 (95% LoA: -0.378–0.254) in IQ200 group. Conclusion: PCCT VNC achieves reliable hepatic steatosis quantification through L/S ratios at 20% reduced dose (IQ155), maintaining high TNC-VNC concordance despite lipid-mediated absolute attenuation biases.