Acquisition and reconstruction parameters

Three acquisitions were performed at 120 kVp and 255 mAs using a retrospective ECG-gated helical acquisition at a chosen heart rate (60 bpm) on each CT system. Tube current modulation was disabled. Beam collimation, rotation time, and pitch factor for the EID-DECT scanner were chosen to imitate clinical practice (Table ​(Table11).

Acquisition and reconstruction parameters used on the dual-layer energy-integrating detector dual-energy CT systems (EID-DECT) (IQon CT, CT7500) and spectral photon-counting CT (SPCCT) systems

HRB (high-resolution standard filter) and CB (cardiac standard filter) were used for high-resolution (HR) imaging in combination with 0.67 mm slice thickness. Detailed 2 was used for ultra-high-resolution (UHR) imaging in combination with 0.43 mm slice thickness. Slice thickness was set at 0.43 mm contiguous for the SPCCT, i.e., adapted to the reconstructed in-plane pixel size for isotropic voxel size, and at 0.67 mm for matching the minimal slice thickness available with EID-DECT. Note that slice increment was set at half of the slice thickness for EID-DECT images such as performed in clinical practice. Slice increment was similar for SPCCT images enabled by thinner slice thickness

^†Detailed 2 and HRB have the same cut-off frequency at 14.6-line pairs, but Detailed 2 filter has higher frequency at 50% and 20% modulation transfer function values

^‡CB filter cut-off is at 12-line pairs per centimeter

Data were reconstructed during the mid-diastolic phase (78% of the R-R interval) of the cardiac cycle. For EID-DECT, raw data were reconstructed according to clinical practice using high-resolution (HR) parameters, with level 1 of the 7 available levels in the spectral reconstruction algorithm, the highest-frequency kernel available CB (Cardiac Standard) and the minimal slice thickness achievable (0.67 mm), corresponding to HR–EID-DECT images. For SPCCT, raw data were reconstructed with level 6 of an adapted hybrid iterative reconstruction algorithm (iDose⁴). Two different-resolution images were reconstructed with HR-SPCCT images using an HRB kernel and a 0.67-mm slice thickness, and ultra-high-resolution (UHR) SPCCT images were reconstructed using a Detailed-2 kernel and a 0.43-mm slice thickness.

To compare the performance of the SPCCT VMIs with conventional acquisitions, an additional reconstruction was performed with the same parameters. This comparison was not made for EID-DLCT because different reconstruction kernels (CB vs. XCB) and reconstruction algorithms (a spectral reconstruction algorithm vs. the iDose⁴ algorithm) were used for spectral and conventional acquisitions.

For all images and both phantoms, the field of view was set at 220 mm and a standard matrix size of 512 × 512 pixels was used. The other acquisition and reconstruction parameters used for both CT systems are depicted in Table ​Table11.