Incidence of Unreliable Automated Computed Tomography Perfusion Maps Zachary Bulwa, MD,* Hormuzdiyar Dasenbrock, MD,† Nicholas Osteraas, MD,‡ Laurel Cherian, MD,‡ R. Webster Crowley, MD,† and Michael Chen, MD† Introduction: Despite the increasing national adoption of automated computed tomography perfusion (CTP) to select thrombectomy patients 6 hours after last known well, reliability issues have been anecdotally reported. Unreliable diagnostic tests add time and confusion to a process that requires efficiency. Our study aims to critically assess an automated CTP program in a contemporary cohort of patients presenting with large vessel occlusion (LVO) in the extended time window by eval- uating the rate of unreliable automated CTP maps and whether this influences clini- cal outcomes. Methods: A retrospective review of consecutive thrombectomy candidates undergoing CTP imaging in the extended time window was performed. All automated CTP maps using RAPID software (iSchemaView, Menlo Park, CA) were assessed for reliability. Clinical outcomes were compared between patients with and without reliable RAPID reports. Results: Ninety-nine consecutive throm- bectomy candidates underwent automated CTP imaging from February 2017 to December 2018. Of these, 78 (79%) had LVO determined by CT angiographyand were included in the study population. Automated CTP maps were unreliable in 13% of cases as a result of motion artifact (n = 3) and contrast bolus flow issues (n = 7). Heart failure was more frequent in patients with unreliable studies. Clinical outcomes did not significantly differ between patients with and without unreliable studies. Conclusions: Thirteen percent of CTP maps generated by automated soft- ware were unreliable, with an increased frequency among patients with heart fail- ure. Given the rate of unreliable automated CTP maps, further studies are warranted to not only establish the true necessity of currently available CTP soft- ware, but also more reliable methods to select patients for thrombectomy present- ing in the extended time window. Key Words: CT perfusion—automated neuroimaging—acute ischemic stroke— extended time window—mechanical thrombectomy © 2019 Elsevier Inc. All rights reserved. Introduction Randomized clinical studies have demonstrated that select patients with anterior large vessel occlusions (LVO) may benefit from thrombectomy after 6 hours from last known well or with unknown time of symptom onset. 1-2 The automated computed tomography perfusion (CTP) software used in these trials (RAPID, iSchemaView, Menlo Park, CA) estimated an infarct core and salvage- able penumbral cerebral parenchymal tissue. While these studies were used to prove the effectiveness of thrombec- tomy in the extended time window, concerns exist with regard to its generalizability. Published study results may be susceptible to underreporting the occurrence of patients who did not fit inclusion criteria, including those with unreliable studies. Additionally, automated software must be purchased, is not widely available, and can at times be difficult to interpret. Recent guidelines have incorporated automated CTP imaging to select stroke patients for endovascular thrombectomy when presenting From the *University of Chicago, Department of Neurology, Chi- cago, Illinois; †Rush University, Department of Neurosurgery, Chi- cago, Illinois; and ‡Rush University, Department of Neurology, Chicago, Illinois. Received July 4, 2019; revision received September 26, 2019; accepted October 5, 2019. Funding: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Address correspondence to University of Chicago Medical Center, Department of Neurology, 5841 South Maryland Avenue, Chicago, IL 60637 E-mail: zachary.bulwa@uchospitals.edu. 1052-3057/$ - see front matter © 2019 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jstrokecerebrovasdis.2019.104471 Journal of Stroke and Cerebrovascular Diseases, Vol. &&, No. && (&&), 2019: 104471 1 ARTICLE IN PRESS