Diagnostic Performance of AI-enabled Plaque Quantification from Coronary CT Angiography Compared with Intravascular Ultrasound

Purpose To assess the diagnostic performance of a coronary CT angiography (CCTA) artificial intelligence (AI)-enabled tool (AI-QCPA; HeartFlow) to quantify plaque volume, as compared with intravascular US (IVUS). Materials and Methods A retrospective subanalysis of a single-center prospective registry study was conducted in participants with ST-elevation myocardial infarction treated with primary percutaneous coronary intervention of the culprit vessel. Participants with greater than 50% stenosis in nonculprit vessels underwent CCTA, invasive coronary angiography, and IVUS of nonculprit lesion(s) between 2 and 40 days after primary percutaneous coronary intervention. Comparisons of plaque volumes obtained using AI-QCPA (HeartFlow) and IVUS were assessed using Spearman rank correlation (ρ) and Bland-Altman analysis. Results Thirty-three participants (mean age, 59.1 years ± 8.8 [SD]; 27 [82%] male and six [18%] female participants) and 67 vessels were included for analysis. There was strong agreement between AI-QCPA and IVUS in vessel (ρ = 0.94) and lumen volumes (ρ = 0.97). High agreement between AI-QCPA and IVUS was also found for total plaque volume (ρ = 0.92), noncalcified plaque (ρ = 0.91), and calcified plaque (ρ = 0.87). Bland-Altman analysis demonstrated AI-QCPA underestimated total plaque volume (-9.4 mm³) and calcified plaque (-11.4 mm³) and overestimated for noncalcified plaque (2.0 mm³) when compared with IVUS. Conclusion An AI-enabled automated plaque quantification tool for CCTA had high agreement with IVUS for quantifying plaque volume and characterizing plaque. Keywords: Coronary Plaque, Intravascular US, Coronary CT Angiography, Artificial Intelligence Supplemental material is available for this article. ClinicalTrials.gov registration no. NCT02926755 © RSNA, 2024.