Improvement of Shunt Detection with Arterial Spin-Labeling MR Imaging in Follow-Up of a Treated Cerebral Arteriovenous Malformation

Abstract

Objective: To identify and evaluate the pitfalls of using arterial spin labeling (ASL) to diagnose cerebral arteriovenous malformation (AVM) and improve its accuracy in clinical practice. Materials and Methods: A retrospective study of 54 patients with cerebral AVM was performed. Each patient underwent magnetic resonance imaging (MRI) studies, including pseudo-continuous ASL (pCASL) and conventional cerebral angiography with a digital subtraction angiography (DSA) during a 3-year period. The consensus of the results of the imaging studies was used to evaluate the diagnostic performance of the MRI technique, with DSA used as the gold standard. Results: Diagnostic accuracy was 94.4% when a positive arteriovenous shunt (AVS) was defined as a high ASL signal in the venous structure or nidus, compared to magnetic resonance angiography (MRA). The misinterpretations of the ASL images of cerebral AVM were due to an arterial transit artifact (ATA) associated with a residual AVS. The other pitfalls were micro-AVMs and very slow-flow AVSs. Conclusion: To improve the accuracy of ASL in evaluating an AVS in cerebral AVM, it is mandatory to detect high ASL signals in venous structures by comparing with MRA.