False-positive diagnosis of cerebral aneurysms using MR angiography: location, anatomic cause, and added value of source image data

Abstract

Purpose: To investigate the anatomic causes of false-positive unruptured aneurysms (FPUIAs) and the added value of source images (SIs) in magnetic resonance angiography (MRA)-based UIA diagnosis.

Methods: The MRA images of 59 patients with 63 FPUIAs and 113 patients with 127 aneurysms were retrospectively reviewed. Two neurointerventionists reviewed MRA- maximum intensity projection (MIP) and conventional angiographic images of patients with FPUIAs, and determined the anatomical causes of FPUIAs by location. They also reviewed both MIP images alone (MIP mode) and additional SI together with MIP (MIP+SI mode) and rated aneurysm probability separately. Receiver operating characteristic (ROC) analysis was performed to compare diagnostic performance of both image modes.

Results: FPUIAs were most commonly found at the internal carotid artery (ICA)-posterior communicating artery (Pcom) (36%). False-positive results at the ICA-Pcom and ICA-anterior choroidal artery resulted from the presence of infundibuli in 28 (97%) and six (100%), respectively. An arterial loop was the leading cause of FPUIAs throughout all locations of the anterior cerebral artery and middle cerebral artery except the anterior communicating artery, where fenestration was found in six (60%) cases. The areas under the ROC curves of the two image modes were not significantly different (0.887 versus 0.925; p=0.103). Addition of the SIs did not cause a significant change in sensitivity (88.2 versus 83.5%; p=0.21), whereas it led to a significant increase in specificity (74.6 versus 95.2%; p=0.0002).

Conclusions: MRA-based FPUIAs are mostly attributable to infundibuli and arterial loops. Although the addition of the SIs appears not to significantly increase the sensitivity of UIA diagnosis, it may significantly improve the specificity.