AI method matches workstation measurement of pericoronary fat attenuation index across five CT scanners

AI measurement of pericoronary fat attenuation index matched workstation values in 1,125 patients (rho 0.877; mean bias 1.60 HU; 94.84% within limits); performance remained stable across five CT scanners (AUC 0.851 at -70.1 HU).

• Retrospective multi-vendor coronary CT angiography study in 1,125 patients (3,375 coronary arteries) across five CT scanners.
• At a threshold of -70.1 HU, AI-FAI achieved 84.98% accuracy and AUC 0.851, comparable to workstation-FAI.
• AI-FAI showed cross-scanner robustness, with inter-scanner rho 0.829-0.896 and only 3.70-6.06% outliers beyond agreement limits.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.