Influence of operator expertise and coronary luminal segmentation technique on diagnostic performance, precision and reproducibility of reduced-order CT-derived fractional flow reserve technique.

Background: Onsite workstation-based CT-derived Fractional-Flow-Reserve (CT-FFR) is accurate in assessing hemodynamic-significance of coronary stenoses. We aim to describe the influence of operator expertise and luminal-segmentation technique on the diagnostic performance, precision and reproducibility of CT-FFR in identifying hemodynamically-significant stenosis (FFR≤0.8).

Methods: Forty-eight consecutive stable-patients (86 vessels) with suspected CAD underwent research indicated invasive-FFR and 320-detector CT-coronary-angiography (CTA). CT-FFR was derived using reduced-order model on standard desktop-computer. Semi-automated coronary luminal segmentation was performed using focused-technique with manual adjustments at regions of stenosis and calcification or comprehensive-technique with manual adjustments along the entire course of the vessel. CT-FFR analysis was performed using 3 blinded operators; core-laboratory engineer using focused-technique and radiographer and cardiologist using the comprehensive-technique. Diagnostic performance was assessed by area under receiver-operating-curve (AUC). Precision with invasive FFR was determined by Bland-Altman analysis, and reproducibility by intraclass-correlation-coefficient (ICC).

Results: Diagnostic performance was comparable among operators (Engineer: AUC = 0.88, Radiographer 0.84; Cardiologist 0.87; P = 0.59). Coronary luminal-segmentation time was shortest using focused technique (engineer 6:17 ± 2.43 min), compared with comprehensive technique (cardiologist 14.83 ± 7.09, radiographer 24.74 ± 12.65; P < 0.001). Use of focused technique was associated with widest limits of agreement (LOA) with FFR and moderate intra-operator reproducibility (engineer LOA -0.20-0.33; ICC 0.66), when compared with the comprehensive technique which demonstrated narrower LOA and excellent reproducibility [radiographer (LOA -0.17-0.20, ICC = 0.91) and cardiologist (LOA-0.15-0.23, ICC = -0.93)] CONCLUSION: A workstation-based CT-FFR technique was reproducible with high and comparable diagnostic performance among operators with different expertise. A comprehensive luminal segmentation technique was the most time-consuming and associated with the highest reproducibility and precision with FFR.