Aortic endograft surveillance: use of fast-switch kVp dual-energy computed tomography with virtual noncontrast imaging.

OBJECTIVE: To assess endoleak detection and patients' radiation exposure using fast-switch peak kilovoltage (kVp) dual-energy computed tomography (DECT) with virtual noncontrast (VNC) imaging.

METHODS: Institutional review board approved retrospective review of triphasic CTs for endograft follow-up: single-energy true noncontrast (TNC) and dual-energy arterial- and venous-phase postcontrast scans on GE HD-750 64-detector scanners. Iodine-subtracted VNC images generated from dual-energy data. Two radiologists (VNC readers) independently performed 2 reading sessions without TNC images: (1) arterial and VNC and (2) venous and VNC. Interrater agreement, leak detection sensitivity, and dose estimates were calculated.

RESULTS: Original dictations described 24 endoleaks in 78 scans. Virtual noncontrast reader agreement was high (κ = 0.78-0.79). Virtual noncontrast reader ranges for sensitivity and negative predictive value for leak detection were 87.5% to 95.8% and 94.0% to 98.0% in venous phase. Dose reduction estimate was 40% by eliminating one phase and 64% by eliminating 2 phases of imaging.

CONCLUSION: Virtual noncontrast images from fast-switch peak kilovoltage DECT data can substitute for TNC imaging in the postendograft aorta, conferring substantial dose reduction. Eliminating 1 of 2 postcontrast phases further reduces dose, with greater negative predictive value for leak detection in the venous versus the arterial phase. Thus, the use of a monophasic venous-phase DECT with VNC images is suggested for long-term endograft surveillance in stable patients.