> Detailanzeige

Tavakkol, Elham; Kihira, Shingo; McArthur, Mark; Ann, Phoebe; Polson, Jennifer; Zhang, Haoyue; Arnold, Corey; Yoo, Bryan; Linetsky, Michael; Ledbetter, Luke; Salamon, Noriko; Jahan, Reza; Duckwiler, Gary; Saver, Jeffrey L; Liebeskind, David S; Nael, Kambiz

Abstract TP97: Automated Assessment Of DWI-FLAIR Mismatch To Predict Stroke-onset Time

Literatur-
verwaltung

Zur
Merkliste

Lösche von
Merkliste

Per Whatsapp teilen

Schließen

Merkliste



Sie können Bookmarks mittels Listen verwalten, loggen Sie sich dafür bitte in Ihr SLUB Benutzerkonto ein.
  • Medientyp: E-Artikel
  • Titel: Abstract TP97: Automated Assessment Of DWI-FLAIR Mismatch To Predict Stroke-onset Time
  • Beteiligte: Tavakkol, Elham; Kihira, Shingo; McArthur, Mark; Ann, Phoebe; Polson, Jennifer; Zhang, Haoyue; Arnold, Corey; Yoo, Bryan; Linetsky, Michael; Ledbetter, Luke; Salamon, Noriko; Jahan, Reza; Duckwiler, Gary; Saver, Jeffrey L; Liebeskind, David S; Nael, Kambiz
  • Erschienen: Ovid Technologies (Wolters Kluwer Health), 2023
  • Erschienen in: Stroke, 54 (2023) Suppl_1
  • Sprache: Englisch
  • DOI: 10.1161/str.54.suppl_1.tp97
  • ISSN: 0039-2499; 1524-4628
  • Entstehung:
  • Anmerkungen:
  • Beschreibung: Background and Purpose: DWI-FLAIR mismatch is used to determine thrombolytic eligibility in patients with acute ischemic stroke (AIS) when time-to-stroke symptom (TSS) is unknown. Advances in artificial intelligence and machine learning (ML) techniques have shown promising results in automated assessment of TSS. We aimed to use the e-Stroke software (Brainomix, Oxford, UK) to automatically classify DWI-FLAIR mismatch in a cohort of patients with AIS and in a comparative analysis with an expert neuroradiologist. Methods: In this retrospective study, consecutive patients with AIS who had pretreatment MRI, and known TSS were included. DWI-FLAIR mismatch was evaluated by an expert neuroradiologist blinded to TSS. e-Stroke software, which uses a combination of statistical methods and ML classification techniques was used to automatically classify DWI-FLAIR status. TSS was dichotomized using a 4.5 hr cutoff to assess the diagnostic performance of the radiologist vs. e-Stroke software. Results: Thus far 113 patients are included, 53 female, age (mean ± SD): 68.9 ± 17. TSS was 371±249 min (mean ± SD) and the infarct volume was 16.4±26.5 ml (mean ± SD). A total of 60 patients had TSS > 4.5 hrs while 53 had TSS < 4.5 hrs. Diagnostic performance (sensitivity/specificity/accuracy in %) for determination of TSS using DWI-FLAIR mismatch was 75/58.5/66.7 for the radiologist and 43/85/64 for e-Stroke software; without statistical difference (AUC: 0.66 for radiologist, 0.64 for e-Stroke; Delong test, p= 0.59). Subgroup analysis of infarctions of at least 1mL or higher, improved the diagnostic performance for both radiologist (83.8/59.4/71.6) and e-Stroke software (66.7/75.0/71.0), without statistical difference (AUC: 0.71 for radiologist and 0.71 for e-Stroke; Delong test, p= 0.96). Conclusions: Results show that fully automated quantitative approach for DWI-FLAIR mismatch provides similar diagnostic accuracies to expert neuroradiologist in determination of TSS.
  • Zugangsstatus: Freier Zugang