Related Experiment Video
Updated: Jun 1, 2026
Optimized Management of Endovascular Treatment for Acute Ischemic Stroke
Published on: January 18, 2018
Automated Quantitative Assessment of Recanalization in Endovascular Thrombectomy
Mohammad Amin Ebrahimzadeh1, Aditi Deshpande1, Jing Wang1
1From the Department of Mechanical Engineering (M.A.E., A.D., K.L.), University of California, Riverside, USA; Divisions of Vascular Neurology and Neurocritical Care (J.W., L.R.A., O.A., P.T.-F.), Inova Neuroscience and Spine Institute, Department of Medical Education (P.T.-F.), University of Virginia, Inova Fairfax Medical Campus, and Division of Neuroradiology (J.D.M.), Inova Fairfax Medical Campus, Falls Church, VA.
Background:
Acute ischemic stroke (AIS) remains a major cause of death and disability worldwide. Endovascular thrombectomy (EVT) has significantly improved outcomes by restoring cerebral blood flow. The thrombolysis in cerebral infarction (TICI) score, the current standard for recanalization assessment, is a semi-quantitative grading system that is commonly used to evaluate the effectiveness of thrombectomy, but it suffers from inter-rater variability and is poorly correlated with clinical outcomes. Here, we propose a novel and fully automated quantitative TICI (qTICI) score derived from digital subtraction angiography (DSA) during EVT that integrates morphological descriptors, vascular network topology, and perfusion dynamics to provide a numeric quantitative reperfusion metric.
Methods:
Using a prospectively collected cohort of 295 AIS patients who underwent EVT, we developed an automated pipeline to extract vascular morphology and perfusion metrics from post-recanalization DSA images. We integrated selected features to compute qTICI and evaluated its performance against adjudicated extended TICI (eTICI) grades, provided by expert interventional neuroradiologists, to predict clinical outcomes.
Results:
While the qTICI score demonstrated moderate agreement with the eTICI grades (micro-average AUC 0.70), it outperformed eTICI in predicting both early and long-term outcomes among successfully recanalized patients (72-hour NIHSS: AUC 0.60 vs 0.47; 90-day mRS: AUC 0.55 vs 0.46). Combining qTICI with additional clinical variables further enhanced its predictive performance. Multivariable analyses showed a greater contribution of qTICI to outcome predictions, compared with eTICI.
Conclusions:
These findings indicate that automated qTICI offers an objective and enhanced alternative to visual grading with modest improvements in prognostic accuracy for AIS outcomes. Prospective, multi-center validation is warranted to standardize reperfusion assessment and support integration of qTICI into clinical trials and stroke care.
