Object size-dependent bias in aortic valve calcium quantification: polychromatic images vs virtual monochromatic images

Objectives To compare aortic valve calcification (AVC) quantifications between polychromatic images (PI) and 70-keV virtual monochromatic images (VMI) obtained from photon-counting detector CT (PCD-CT), and to assess the influence of object size on the quantifications. Materials and methods This retrospective study included participants who underwent PCD-CT for aortic stenosis (AS) assessment or intervention planning between March 2023 and August 2025. AVC was quantified using aortic valve Agatston score (AVAS), calcium volume (AVCV), and calcium mass (AVCM) from both PI and 70-keV VMI. Wilcoxon signed-rank test assessed differences, and Spearman’s rank correlation coefficient and univariable linear regression analyses were used to evaluate associations between the PI/VMI ratio and water-equivalent diameter (WED). Agreement of AVAS-based classification between PI and 70-keV VMI was evaluated using Cohen’s κ. Results A total of 137 participants (median age, 84 years [IQR, 79–88]; 81 women) were included. PI yielded significantly higher values than 70-keV VMI for AVAS, AVCV, and AVCM (all p < 0.001). The PI/VMI ratio showed a strong negative correlation with WED, particularly for AVCM (ρ = -0.73). In univariable regression analyses, WED showed negative associations with the PI/VMI ratios for AVCM and AVAS. Cohen’s κ for AVAS-based classification was 0.95–0.96. Conclusion Compared with VMI-derived AVC quantification, PI-derived AVC quantification may be influenced by object size-dependent beam hardening. Although this did not lead to substantial AS-severity reclassification, differences in AVC quantification between PI and VMI may warrant consideration. Key Points Question: Do PI and VMI yield different AVC quantifications on PCD-CT? Findings: PI yielded systematically higher AVC volume, mass, and AVAS than 70-keV VMI. These differences may be associated with object size-dependent beam hardening. Relevance statement: Object size-dependent beam hardening may contribute to systematic differences in AVC quantification between PI and VMI. Although clinical reclassification was limited, reconstruction-specific differences may warrant consideration when applying CT-based thresholds for AS severity.