Evaluation of Virtual Noncontrast Images Obtained from Dual-Energy CTA for Diagnosing Subarachnoid Hemorrhage.

BACKGROUND AND PURPOSE The virtual noncontrast images generated with iodine subtraction from dual-energy CTA images are expected to replace the true noncontrast images for radiation-dose reduction. This study assessed the feasibility of virtual noncontrast images for diagnosing SAH. MATERIALS AND METHODS Eighty-four patients with or without SAH underwent true noncontrast brain CT (the criterion standard for diagnosing SAH). Among them, 37 patients underwent an additional head dual-energy angiography, and the other patients underwent head and neck dual-energy angiography. Virtual noncontrast images were produced on a dedicated dual-energy postprocessing workstation and reconstructed in orientation and section width identical to those in true noncontrast images. The findings on the virtual noncontrast and true noncontrast images were compared at both the individual level and the lesion level. Image noise of the virtual noncontrast and true noncontrast images was also measured and compared. The volume CT dose index and dose-length product were recorded for the radiation-dose analysis. RESULTS The sensitivity, specificity, positive predictive value, and negative predictive value of virtual noncontrast images at the individual level and the lesion level were 94.5%, 100%, 100%, 90.6% and 86.7%, 96.9%, 91.8%, 94.8%, respectively. The agreement in the diagnosis of SAH on true noncontrast and virtual noncontrast images reached 92.3% at the individual level and 85.1% at the lesion level. The virtual noncontrast images showed a higher image noise level. The volume CT dose index and dose-length product were obviously reduced without the true noncontrast brain CT scan. CONCLUSIONS Virtual noncontrast images are a reliable tool for diagnosing SAH, with the advantage of reducing the radiation dose.