Virtual unenhanced imaging of the liver with third-generation dual-source dual-energy CT and advanced modeled iterative reconstruction

Objectives To compare image quality and diagnostic accuracy for the detection of liver lesions of virtual unenhanced (VU) images based on third-generation dual-source dual- energy computed tomography (DECT) compared to conventional unenhanced (CU) images. Methods Thirty patients underwent triphasic abdominal CT consisting of single-energy CU (120 kV, 147 ref.mAs) and dual-energy CT arterial and portal-venous phase acquisitions (100/Sn150 kV, 180/90 ref.mAs). VU images were generated from arterial (AVU) and portal venous (PVU) phases. CU, AVU and PVU datasets were reconstructed. Quantitative image quality analysis was performed and two abdominal radiologists independently analyzed all datasets to evaluate image quality and identify liver lesions. Radiation dose was recorded and potential radiation dose reduction was estimated. Results Image quality was rated diagnostic in 100% of the VU datasets. The mean subjective image quality of the CU datasets was higher than that of VU images (p < 0.0001). No significant difference was observed in the mean attenuation values of the liver parenchyma (p > 0.99) and hypoattenuating liver lesions (p ≥ 0.21) between CU, AVU and PVU. However, a significant reduction in the attenuation values of calcified lesions (p < 0.0001), metallic clips (p < 0.0001) and gallstones (p ≤ 0.047) was observed in the AVU and PVU images compared with CU images. A total of 122 liver lesions were found in 25 patients. VU images were more sensitive than CU images for detection of small hypoattenuating liver lesions (≤1 cm). However, CU images were more sensitive than VU for calcified liver lesions. The mean radiation dose reduction achievable by avoiding the unenhanced acquisition was 32.9% ± 1.1% (p < 0.01). Conclusions Third-generation DSCT VU images of the liver provide diagnostic image quality and improve small (≤1 cm) liver lesion detection; however calcified liver lesions can be missed due to complete subtraction