Highly efficient neuroimaging for endovascular mechanical thrombectomy patient triage


Dylan N Wolman and Jeremy J Heit write in NeuroNews about the importance of neuroimaging and the role it plays in patient evaluation.

Dylan Wolman

Acute ischaemic stroke secondary to large vessel occlusion (LVO) of the anterior circulation is a devastating condition, with an approximately 70% risk of death or serious disability if untreated or treated by medical therapy alone.1,2 Recent unprecedented advances in acute ischaemic stroke treatment by endovascular mechanical thrombectomy have been shown to reduce significantly the morbidity and mortality of these severe strokes.3-5 Following the success of these endovascular trials, the stroke neurology and neurointerventional communities are treating many more patients with acute ischaemic stroke due to LVO,and their attention is increasingly turning toward effective systems of care and the most expeditious neuroimaging evaluation of endovascular mechanical thrombectomy candidates. Rapid and accurate identification of endovascular mechanical thrombectomy candidates both at comprehensive stroke centres and in smaller referring hospitals is essential to providing timely and appropriate care to acute ischaemic stroke patients.

Jeremy Heit

The neuroimaging evaluation of acute ischaemic stroke patients has advanced in parallel to endovascular mechanical thrombectomy. Patients most likely to benefit from endovascular mechanical thrombectomy have a small core infarction, salvageable tissue at risk of infarction (Penumbra), and an LVO.1-8 These patient characteristics should be identified on non-invasive computerised tomography (CT) and magnetic resonance imaging (MRI) studies prior to performing endovascular mechanical thrombectomy to ensure that acute ischaemic stroke patients are being treated appropriately.9 There is wide variation in imaging protocols designed to answer these questions, but more comprehensive protocols with perfusion imaging have shown a larger treatment effect when compared to protocols that do not incorporate perfusion imaging.5-13

In these more comprehensive imaging protocols, the size of the core infarction is identified based upon diffusion-weighted imaging (DWI) MRI or CT perfusion estimates of core infarction, the presence and size of salvageable tissue is determined by CT or MRI perfusion imaging, and the presence of an LVO is determined by CT or MR angiography.9 However, the large amount of information provided by these comprehensive imaging protocols must be balanced with a several minute time delay that is required to perform perfusion imaging.

The ideal non-invasive imaging protocol to determine endovascular mechanical thrombectomy candidacy would be fast, accurately identify the presence of a small core infarction, salvageable tissue at risk of infarction and LVO presence, and easily performed at both comprehensive and referring hospitals. At Stanford University Medical Centre, USA, we most commonly use MRI for endovascular mechanical thrombectomy triage. Our simplest and most rapid acute ischaemic stroke protocol includes: DWI, gradient-echo imaging (to exclude cerebral haemorrhage), time-of-flight MR angiography, and MR perfusion imaging. This protocol has a total imaging time of five minutes and 10 seconds and answers all questions necessary for appropriate endovascular mechanical thrombectomy triage.14 However, we thought that this short protocol could be further streamlined.

Perfusion imaging provides excellent delineation of salvageable tissue, but it also provides important information regarding the presence and location of an LVO. Neurointerventionalists can leverage their detailed knowledge of cerebral vascular anatomy to infer the location of an LVO based upon the pattern of a perfusion deficit. We hypothesised that perfusion imaging would provide the same information as CT or MR angiography regarding the presence of an LVO, and we leveraged this concept in a recent publication.14 We retrospectively determined whether patients with acute ischaemic stroke symptoms could be accurately triaged to endovascular mechanical thrombectomy based only on DWI and MR perfusion imaging alone.14

Three readers reviewed DWI and MR perfusion images from 219 patients with acute ischaemic stroke symptoms, 73 of whom underwent endovascular mechanical thrombectomy for acute ischaemic stroke due to LVO. The readers identified the presence of a small core infarction on DWI, the presence of salvageable tissue on MR perfusion imaging, and LVO presence and location on MR perfusion imaging. Patients with these three characteristics were deemed endovascular mechanical thrombectomy candidates by the readers, and results were compared to those made at the time of patient presentation. We found DWI and MR perfusion alone to have a remarkable 96% sensitivity and 98% specificity for accurate endovascular mechanical thrombectomy triage. Additionally, readers accurately localised LVOs to the ICA/M1 segment in 96% of cases and the M2 segment in 88% of cases, while correctly identifying patients with an LVO in 99% of cases. We concluded that DWI and MR perfusion alone is highly accurate, sensitive, and specific for appropriate endovascular mechanical thrombectomy triage, LVO identification, and LVO localisation.

Our results strongly suggest that conventional vessel imaging with MR or CT angiography may be confidently replaced with perfusion imaging in nearly all patients. The omission of non-invasive angiography offers a modest time savings of approximately 2.5 minutes, which may be beneficial in the care of acute ischaemic stroke patients. More importantly, however, is the applicability of our results to acute ischaemic stroke patient triage from transferring hospitals to comprehensive stroke centres, which has important implications for stroke systems of care. Our study was performed with the gold standard MRI, but it logically follows that similarly accurate patient triage to endovascular mechanical thrombectomy and LVO detection and localisation may be performed with CT perfusion techniques.

Consider the following example, which is based upon the system we have instituted at Stanford with several of our referring hospitals. A community hospital performs CT perfusion in an acute ischaemic stroke patient with symptoms that are suspicious for an LVO. These images are processed automatically using RAPID (iSchemaView), and the summary images from the CT perfusion study are automatically forwarded to the comprehensive stroke team members as an email. These summary images provide a volumetric estimation of core infarction and of the salvageable tissue at risk as well as the identification of an LVO, as shown in our study. Furthermore, these images can be easily interpreted on a mobile phone. The community hospital physician can then call the stroke team to discuss transfer of the patient, and the comprehensive stroke team members can determine endovascular mechanical thrombectomy eligibility with a high degree of accuracy. Moreover, futile patient transfers due to the absence of an LVO, a large core infarction (>70 ml), or a volumetric match between the core infarction and salvageable penumbra are avoided. Currently, up to 72% of acute ischaemic stroke patients transferred for possible endovascular mechanical thrombectomy do not undergo treatment secondary to futile transfer, which underscores the need to simplify and improve acute ischaemic stroke patient triage. Our early experience with this stroke system of triage has been very successful, and we are working to expand our programme to additional community hospitals to optimise the care of acute ischaemic stroke patients.

Many comprehensive stroke centres are currently struggling with a sudden and marked increase in the volume of acute ischaemic stroke patients requiring evaluation for possible endovascular mechanical thrombectomy following the positive results of the DAWN and DEFUSE 3 trials, which have extended the endovascular mechanical thrombectomy treatment window to 16-24 hours.4, 5 Both of these studies used CT perfusion to estimate the core infarction in the majority of patients (the other patients underwent evaluation with MRI), and it is likely that perfusion imaging will become increasingly used in the evaluation of acute ischaemic stroke patients, particularly in late time windows. The results of our study provide a helpful roadmap to comprehensive centres as they evaluate their own systems of care following the successful DAWN and DEFUSE 3 trials.

The remarkable advancements in acute ischaemic stroke treatment by endovascular mechanical thrombectomy now require pre-endovascular mechanical thrombectomy imaging triage to evolve toward protocols that are efficient and provide the best information for safe, accurate, and expeditious patient treatment. Our study is the first to test the hypothesis that perfusion weighted imaging contains adequate angiographic data for endovascular mechanical thrombectomy triage. Although additional prospective studies are necessary to confirm our findings that endovascular mechanical thrombectomy triage may be safely and accurately performed in the absence of CT or MR angiography, we are confident that our results will be helpful to other stroke healthcare systems as they work to provide endovascular mechanical thrombectomy to many more eligible patients.

Dylan N Wolman, Department of Radiology, and Jeremy J. Heit, University Medical School, Stanford, USA

Corresponding Author:

Jeremy J. Heit
Clinical Assistant Professor of Radiology
300 Pasteur Drive, Room S047
Stanford, CA 94305
650-723-6767 (office)
650-498-5374 (fax)
[email protected]


  1. Lima FO et al. Prognosis of untreated strokes due to anterior circulation proximal intracranial arterial occlusions detected by use of computed tomography angiography. JAMA Neurol. 2014;71(2):151-7.
  2. Bhatia R, et al. Low rates of acute recanalization with intravenous recombinant tissue plasminogen activator in ischaemic stroke: real-world experience and a call for action. Stroke. 2010;41(10):2254-8. .
  3. Saver JL et al. Time to Treatment With Endovascular Thrombectomy and Outcomes From Ischaemic Stroke: A Metaanalysis. JAMA. 2016;316(12):1279-88.
  4. Nogueira RG et al. Thrombectomy 6 to 24 Hours after Stroke with a Mismatch between Deficit and Infarct. NEJM. 2018;378(1):11-21.
  5. Albers GW et al. Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging. NEJM. 2018;378(8):708- 18. Epub 2018/01/25.
  6. Saver JL et al. Stent-retriever thrombectomy after intravenous t-PA vs. t-PA alone in stroke. NEJM. 2015;372(24):2285-95.
  7. Campbell BC et al. Endovascular therapy for ischaemic stroke with perfusion-imaging selection. NEJM. 2015;372(11):1009-18.
  8. Albers GW. Late Window Paradox. Stroke. 2018;49(3):768-71.
  9. Heit JJ, Wintermark M. Imaging selection for reperfusion therapy in acute ischaemic stroke. Curr Treat Options Neurol.2015;17(2):332.
  10. Patel VP, Heit JJ. Ischaemic Stroke Treatment Trials: Neuroimaging Advancements and Implications. Top Magn Reson Imaging. 2017;26(3):133-9.
  11. Berkhemer OA et al. A randomized trial of intraarterial treatment for acute ischaemic stroke. NEJM. 2015;372(1):11-20.
  12. Jovin TG et al. Thrombectomy within 8 hours after symptom onset in ischaemic stroke. NEJM. 2015;372(24):2296-306.
  13. Goyal M et al. Randomized assessment of rapid endovascular treatment of ischaemic stroke. NEJM. 2015;372(11):1019-30.
  14. Wolman DN et al. Can diffusion- and perfusion-weighted imaging alone accurately triage anterior circulation acute ischaemic stroke patients to endovascular therapy? Journal of Neurointerventional Surgery. 2018. Epub 2018/03/21.


Please enter your comment!
Please enter your name here