To evaluate the contribution of high-energy virtual monochromatic images (VMIs) and mono-bin images of a clinical prototype spectral photon-counting CT (SPCCT) for metallic artefacts reduction on cardiac mechanical valves.

Two mechanical valves (SJM, Abott) of respectively 25 and 17mm diameter were scanned in a tube containing iodine placed in a thorax phantom using a clinical retrospective ECG-gated coronary CTA protocol (120 kVp, 255 mAs) with SPCCT (Philips Healthcare) with and without an extension ring. Conventional, VMIs from 40keV to 120 keV (for all acquisitions) and mono-bin (for one acquisition) images were reconstructed using a high-frequency kernel and a 1024x1024-matrix size (0.21x0.21mm voxel size). Blooming artefact was assessed by drawing a lumen profile along the external rings and measuring the area under the curve (AUC) of the profile and the kurtosis which describes the distribution of this profile. Beam hardening was measured by placing regions of interest (ROIs) around the rings and calculating the percentage of pixels with a values < 3-times the standard deviation in iodine. This analysis was performed with an in-house built tool.

AUC of the profile calculated on VMIs showed progressively decreasing values going from lower to higher keV energies. AUC at 40keV was higher than on conventional images. AUC was significantly lower than conventional images starting from 100keV (24.5±7.0 vs 27.6±7.4, p=0.004) and higher energies. The same pattern was found for beam hardening artefacts. The lowest of the high energies yielding significantly lower values compared to conventional images was 90keV (7.9±5.4 vs 13.8±7.9, p=0.03). For the one tested valve, mono-bin images 3 and 4 showed lower AUC and beam hardening values than at these high keV (although no statistical analysis could be performed). Kurtosis showed a more various behaviour at different VMI energies, possibly due to the complex geometrical structure of the two valvular rings.

Using high-energy VMIs, SPCCT enabled a significant reduction of blooming and beam hardening artefacts compared to conventional images for metallic valve.  Mono-bin images are also a promising tool to reduce these artefacts.