Accuracy of Vessel-Encoded Pseudo-Continuous Arterial Spin Labeling in Identification of Feeding Arteries in Patients with Intracranial Arteriovenous Malformation

Songlin Yu¹, Rui Wang², Rong Wang¹, Shuo Wang¹, Yuqiang Yao³, Dong Zhang¹, Yuanli Zhao¹, Zhentao Zuo², Rong Xue², Danny J.J. Wang⁴, Jizong Zhao¹

In order to evaluate the diagnostic accuracy of VE-PCASL in identifying feeding arteries of intracranial AVM, sixteen AVM patients were examined with VE-PC ASL. Supply fraction of each feeding artery to the AVM was calculated. ROC curves were calculated to evaluate the diagnostic accuracy of VE-PCASL for identifying feeding arteries of AVMs, using DSA as the gold standard. Results are that for VE-PCASL with standard labeling efficiencies, the AUC was 0.935. The optimal cut-off of supply fraction for identifying feeding arteries was 15.17% and the resulting sensitivity was 83.3% and specificity was 91.7%. For VE-PCASL with custom labeling efficiencies, the AUC was 0.956. The optimal cut-off of supply fraction was 11.73% which yielded 88.9% sensitivity and 91.7% specificity. We conclude VE-PCASL with either standard or custom labeling efficiencies offers a high level of diagnostic accuracy compared to DSA for identifying feeding arteries.