Influence of X-ray Tube Voltage on Computed Tomography Values and Contrast-Noise Ratios in Virtual Monoenergetic Images in routine clinical Photon Counting Detector Computed Tomography

PURPOSE: Conventional energy integrating detector (EID) CT provides polyenergetic images where CT values depend on the tube voltage (kV), hampering quantitative analysis across different CT scans. To mitigate this problem, spectral CT such as photon-counting detector (PCD) CT offers virtual monoenergetic images (VMI) where CT values depend only on the keV level but, ideally, not on the tube voltage. The purpose of this study was to investigate the kV dependence of VMI CT values in PCD-CT across a wide range of tissue types, keV levels, and kV settings for various phantom sizes, and to determine which kV settings yield the optimal contrast-to-noise ratio (CNR) for a given VMI energy and phantom size. The phantom study was complemented with an exploratory patient study. MATERIALS AND METHODS: A spectral abdomen phantom (QSA-543; QRM) with 2 optional fat rings was scanned on a PCD-CT (NAEOTOM Alpha; Siemens Healthineers). The phantom contained rods with 2/5/10/15 mg/mL iodine (QRM); muscle/fat/liver equivalent tissue; and 400 mg/mL calcium (Gammex Sun Nuclear). A standard abdomen protocol was used at tube voltages of 70/90/120/140 kV. Polyenergetic images and VMIs (range: 40 to 100 keV in 5 keV increments) were reconstructed. For each VMI energy (keV), the percentage deviation ΔCT(%) of the mean CT value across the different kV settings was determined for each rod using region-of-interest (ROI) analysis. A ΔCT of 10% was considered the clinically acceptable upper limit. For each kV setting, the iodine CNR was calculated and normalized to the radiation dose (CTDIvol) to obtain CNRD. Higher CNRD indicated higher radiation dose efficiency. Edge spread functions (EFS) and noise power spectra (NPS) were evaluated for various kV settings. In the retrospective patient study, a standard cerebral CT angiography (CTA) protocol was applied at either 140 kV (n = 42) or 90 kV (n = 25), with VMI reconstructed at 55 keV. Both the ΔCT(%) and the CNRD relative to the ventricles were determined for various cerebral arteries. The 2 groups were compared using unpaired t tests. RESULTS: For all iodine and calcium rods and all phantom sizes, CT values remained stable within the predefined 10% tolerance across all kV settings in the VMI energy range of 40 to 70 keV, with ΔCT ≤8.4%. At 100 keV, ΔCT increased for low iodine concentration, reaching up to 36%, corresponding to a deviation of 19 HU from the mean CT value of 54 HU for 5 mg/mL iodine. Adipose tissue and liver exhibited ΔCT ≤10% at 45 to 70 keV. At 40 keV, deviations were larger but remained small in absolute terms (11.0 HU for liver and 19.6 HU for adipose tissue in the large phantom). For the small phantom and 40-60 keV, CNRD was 10% to 13% higher for 70/90 kV compared with 120/140 kV. EFS and NPS were consistent across kV settings. This advantage was lost at higher keV and larger phantoms. In patient VMIs at 55 keV, CNRD was 17% to 29% higher at 90 kV than at 140 kV, albeit generally not statistically significant. CONCLUSIONS: In VMIs acquired with PCD-CT, CT values were stable within the predefined 10% tolerance for all phantom sizes and tube voltages in the energy range of 45 to 70 keV, which is relevant for contrast-enhanced CT. This represents a significant step towards the meaningful comparison of CT values across different CT scans. At low keV and small phantom sizes, scans at 70 or 90 kV demonstrated greater radiation dose efficiency compared with scans at 120 or 140 kV. Therefore, for routine contrast-enhanced CT examinations of small patients or small body parts (eg, CTA of the head), scanning at low tube voltage may be considered if no spectral results other than VMIs are required.