The importance of spectral separation: an assessment of dual-energy spectral separation for quantitative ability and dose efficiency
Bernhard Krauss,Katharine L Grant,Bernhard T Schmidt,Thomas G Flohr
DOI: https://doi.org/10.1097/RLI.0000000000000109
Abstract:Introduction: One method to acquire dual-energy (DE) computed tomography (CT) data is to perform CT scans at 2 different x-ray tube voltages, typically 80 and 140 kV, either as 2 separate scans, by means of rapid kV switching, or with the use of 2 x-ray sources as in dual-source CT (DSCT) systems. In DSCT, it is possible to improve spectral separation with tin prefiltration (Sn) of the high-kV beam. Recently, x-ray tube voltages beyond the established range of 80 to 140 kV were commercially introduced, which enable additional voltage combinations for DE acquisitions, such as 80/150 Sn or 90/150 Sn kV. Here, we investigate the DE performance of several x-ray tube voltages and prefilter combinations on 2 DSCT scanners and the impact of the spectra on quantitative analysis and dose efficiency. Materials and methods: Circular phantoms of different sizes (10-40 cm in diameter) equipped with cylindrical inserts containing water and diluted iodine contrast agent (14.5 mg/cm) were scanned using 2 different DSCT systems (SOMATOM Definition Flash and SOMATOM Force; Siemens AG, Forchheim, Germany). Five x-ray tube voltage combinations (80/140 Sn, 100/140 Sn, 80/150 Sn, 90/150 Sn, and 100/150 Sn kV) were investigated, and the results were compared with the previous standard acquisition technique (80/140 kV). As an example, 80/140 Sn kV means that 1 x-ray tube of the DSCT system was operated at 80 kV, whereas the other was operated at 140 kV with additional tin prefiltration (Sn). Dose values in terms of computed tomography dose index (CTDIvol) were kept constant between the different voltage combinations but adjusted with regard to object size according to automatic exposure control recommendations. Reconstructed images were processed using linear blending of the low- and high-kV CT images to combined images, as well as 3-material decomposition techniques to generate virtual noncontrast (VNC) images and iodine images. Contrast and pixel noise were evaluated, as well as DE ratios, which are defined as the CT value at low kV divided by the CT value at high kV. Results: For the 10-, 20-, 30-, and 40-cm phantom, dose values in terms of CTDIvol were 1.2, 2.6, 7.3, and 21.6 mGy, respectively. In the combined images, those obtained with tin filtration showed lower noise values at similar iodine enhancement levels than did images obtained without tin filtration. The largest differences in noise were observed for the larger phantoms, in particular the 40-cm phantom. Dual-energy ratios for iodine increased with decreasing voltages of the low-kV beam and with increasing voltages of the high-kV beam, and they increased when tin prefiltration was added. In case of the 20-cm phantom, DE ratios ranged from 2.0 at 80/140 kV to 3.4 at 80/150 Sn kV. The noise level of the VNC images was strongly correlated with the inverse of the DE ratio. Irrespective of the phantom size, the lowest noise values were measured for 80/150 Sn kV. Discussion: Dual-source CT systems enable DE data to be acquired using a variety of voltage combinations. Combined (or mixed) DE images provide an image impression similar to standard 120 kV images, yet the noise level depends on the DE voltage combination that is selected. Noise in decomposed VNC images is strongly influenced by the DE ratio, and it improves substantially with tin filtration of the high-voltage beam.