We retrospectively analysed 30 angio-DECT performed between January and December 2018.
CT acquisitions
CT acquisitions were performed using a dual-layer detector CT with retrospective ECG gating.
Standard coronary CT scan parameters were employed (tube potential, 120 kVp; tube current, 220 mAs; portal revolution time of 0.27s; automatic exposure control (angular and longitudinal); detector collimation of 64x0.625mm). Bolus tracking was used to start the acquisition.
For all patients the same contrast material (Iomeron 400) was employed. The injection protocol was: 70 mL in patients 80 kg, maximum 90 mL.
CT reconstructions
For each exam different series were reconstructed: conventional CT equivalent, monoenergy at 70, 60, 50 and 40keV. All reconstructions had a slice thickness of <1mm.
Manual aortic measurements
Measurements were independently performed by two radiologists with 7 and 2 years experience in cardiovascular radiology, respectively.
Manual measurements of the descending aorta were obtained at the last slice where the inferior left pulmonary vein was visible on a plane perpendicular to the aorta’s longitudinal axis on double oblique reconstructions. At this level the maximum inner-edge to inner-edge diameter was measured for conventional, 70keV and 40keV reconstructions independently and at different times.
Window level was adjusted to the observers’ preference while other parameters, such as the zoom, were fixed.
Automatic aortic measurements
A second set of measurements was obtained by using a commercially available automatic measurement software. Each of the conventional, monoE 70keV, monoE 60keV, monoE 50keV, monoE 40keV was treated as a separate dataset and independently uploaded in the software. Small errors in terms of centerline identification were manually corrected. On the contrary, impossibility to define a correct centreline and evident mistakes in the definition of the aortic contours were considered as a software failure. Automatically calculated diameters were noted.
Statistical analysis
Statistical analysis was performed with RStudio version 1.1.456. Bland Altman plots and paired t-test were employed to estimate differences between measurements.