Diagnostic efficacy of dual-energy coronary CT angiography for the identification of myocardial focal scar
C Pan, Z Yang, L Zeng, X Zhou, T Jia, F Hu, L Lu, P Qin, L Luo, B Chen, S Li, J ChenAbstract
Background
Cardiovascular magnetic resonance late gadolinium enhancement (CMR-LGE) remains the reference standard for the identification of myocardial focal scar (MFS); however, its clinical utility is often constrained by high costs and prolonged acquisition times Dual-energy coronary computed tomography angiography (DECCTA) generates virtual monochromatic images (VMIs) that enhance myocardial characterisation by improving tissue contrast and mitigating beam-hardening artifacts. Nevertheless, the optimal VMIs energy level for the assessment of MFS has yet to be determined.
Purpose
To evaluate the diagnostic efficacy of DECCTA-derived VMIs across various energy levels for the identification and quantification of MFS, utilizing CMR-LGE as the reference standard.
Methods
We prospectively enrolled 47 patients with chronic myocardial infarction and 10 healthy controls who underwent concurrent DECCTA and CMR within a three-day window. DECCTA data were reconstructed into four VMIs levels: 40 keV, 50 keV, 130 keV, and the optimal contrast-to-noise ratio (CNR) keV. Diagnostic performance was evaluated at the global left ventricular (LV), American Heart Association (AHA) segment, slice, and wall levels. Statistical analyses included receiver operating characteristic (ROC) curve analysis for qualitative assessment, as well as Wilcoxon rank-sum tests, error analysis (mean error [ME] and limits of agreement [LOA]), and concordance correlation coefficients (CCC) for quantitative assessment (MFS area percentage).
Results
Analysis of 912 myocardial segments (293 LGE-positive) identified 130 keV as the optimal energy level for qualitative MFS identification. At this level, global LV findings demonstrated no significant difference compared with the reference standard, with high concordance across most AHA segments (except 8, 12, 15, 16), slice levels (except mid-cavity), and wall regions (except lateral). The global LV AUC was 0.63, peaking at 0.70 for mid-cavity slices. For quantitative diagnosis, 50 keV was superior; despite global LV differences across all VMIs, 50 keV yielded the highest accuracy (ME: -0.05; LOA: -0.28 to 0.15; CCC: 0.26) and showed no significant variation across the majority of AHA segments, mid-cavity slices, and anterior/septal walls. Integrating 130 keV qualitative screening with 50 keV quantitative refinement provided a marginal but quantifiable improvement in overall diagnostic precision (ME: -0.056; LOA: -0.27 to 0.16; CCC: 0.29).
Conclusion
DECCTA-derived VMIs at 130 keV and 50 keV provide optimal performance for qualitative identification and quantitative assessment of MFS, respectively. An integrated multi-energy approach yields an incremental enhancement in diagnostic efficacy, suggesting that DECCTA is a promising alternative for comprehensive myocardial scar characterization in clinical practice.