CT angiography in non-traumatic subarachnoid hemorrhage: the importance of arterial attenuation for the detection of intracranial aneurysms.

Abstract

Computed tomography angiography (CTA) is today the primary method for the detection of intracranial aneurysms. The technique has evolved considerably during the last decade, and it is important to establish criteria for high image quality, especially with regard to improving the diagnosis of small aneurysms. To evaluate diagnostic accuracy and image quality by arterial attenuation of CTA in patients with non-traumatic subarachnoid hemorrhage (SAH). Between 2005 and 2011, CTA and digital subtraction angiography (DSA) were performed in 326 patients with non-traumatic SAH. Sensitivity and specificity for aneurysm detection were evaluated per patient, per aneurysm, and per ruptured aneurysm. The image quality of CTA was evaluated by arterial attenuation measurements (mean Hounsfield units [HU]) in the internal carotid artery (ICA). In all, 285 aneurysms in 235 patients were detected by DSA, 19 aneurysms were missed on CTA, and 223 aneurysms were classified as ruptured. In 91 patients, no aneurysm was found. Correct diagnosis with CTA was made in 28 patients with perimesencephalic hemorrhage. Sensitivity and specificity (95% confidence interval) calculated per patient were 91.6% (87.3-94.9) and 87.9% (79.8-93.6), respectively, per aneurysm 93.3% (89.7-95.9) and 88% (79.9-93.6), and per ruptured aneurysm 94.9% (91.3-97.3) and 96.7% (90.7-99.3). Arterial attenuation (in HU) in CTA revealing true positive ruptured aneurysms and true negative aneurysms (mean 535 ± 110 HU) differed significantly (P = 0.02) from false negative ruptured aneurysms (mean 424 ± 30 HU). CTA has high sensitivity and specificity for the detection of ruptured aneurysms. The sensitivity is related to arterial attenuation in the ICA.