Beyond Recanalization: Machine Learning-Based Insights into Postthrombectomy Vascular Morphology in Patients with...

Area of Science:

• Neurology
• Radiology
• Artificial Intelligence

Background:

• Despite successful endovascular therapy (EVT), many stroke patients experience poor outcomes.
• Identifying residual macrovascular perfusion deficits post-EVT is crucial for improving patient outcomes.
• Machine learning (ML) offers a novel approach to analyze complex vascular changes.

Purpose of the Study:

• To investigate the utility of ML-based analysis of vascular changes on MRA after EVT in identifying perfusion deficits.
• To correlate vascular features extracted by ML with early neurological improvement (ENI) in patients with large vessel occlusion (LVO) stroke.
• To explore the potential of ML in guiding adjunctive therapies for stroke patients.

Main Methods:

• Retrospective analysis of 44 patients with anterior circulation LVO stroke and successful recanalization (mTICI 2b/3).
• An ML algorithm extracted vascular features from pre- and 24-hour post-EVT MRA, focusing on ipsilateral arterial branch length increase.
• Perfusion deficits were assessed using PWI, MTT, or distal clot presence; ENI was defined by a significant decrease in NIHSS score within 24 hours.

Main Results:

• ML analysis revealed that patients with distal clot had significantly smaller increases in arterial length (51% vs. 134%, p=0.05).
• Patients achieving ENI demonstrated greater ipsilateral arterial branch length increases (161% vs. 67%, p=0.023) compared to those without ENI.
• Complete reperfusion was achieved in 71% of the studied patients.

Conclusions:

• ML-based vascular analysis of MRA post-EVT can effectively identify macrovascular perfusion deficits.
• Increased arterial length post-EVT, identified by ML, is a significant predictor of early neurological improvement in stroke patients.
• This approach may aid in tailoring adjunctive therapies to optimize outcomes after EVT for LVO stroke.