Clinical applications of virtual, non-contrast head images derived from dual-source, dual-energy cerebrovascular computed tomography angiography

Background: This study set out to evaluate the utility of cerebrovascular virtual non-contrast (VNC) scans. Materials and Methods: Conventional non-contrast (CNC) and dual-energy computed tomography angiography (DE-CTA) head scans were conducted on 100 subjects, of which 46 were normal, 15 had parenchymal hematomas of the brain, 13 had ischemic infarction, 22 had tumors, and 4 had calcified lesions. VNC images were extracted from the DE-CTA head scans by post-processing. The true (or conventional) and VNC images were compared in terms of the mean CT attenuation value and signal-to-noise ratio (SNR) of the cerebral parenchyma, the image quality, the lesion detection sensitivity, and the radiation exposure level. Results: The image qualities of the CNC and VNC scans were (4.95 &plusmn; 0.22) points and (3.94 &plusmn; 0.24) points (t = 31.18, P < 0.05), the mean CT values for the CNC and VNC images were (34.6 &plusmn; 2.44) and (28.6 &plusmn; 5.40) HU (t = 10.126, P < 0.05), the SNRs were (9.45 &plusmn; 1.26) and (6.87 &plusmn; 1.77), and the HU for white matter was (t = 11.859, P<0.05), respectively. The effective radiation doses from the DE-CTA head scans and the conventional non-contrast scans were (8.55 &plusmn; 0.57) mSv and (9.41 &plusmn; 1.00) mSv, respectively. No significant difference in the lesion detection sensitivities was observed between the CNC and VNC scans, except for tiny calcified lesions, which could not be identified by a VNC scan. Conclusion: VNC and contrast-enhanced images could be obtained from DE-CTA head scans and could aid in the diagnosis of cerebral lesions. The radiation dose from the VNC scan was less than that from the CNC scan.