SIR 2024
Neurointerventional Radiology
Adnan Khan, BS
Medical Student
Texas A&M School of Engineering Medicine
Financial relationships: Full list of relationships is listed on the CME information page.
James Zhang, BS, MS
Medical Student
Texas A&M School of Engineering Medicine
Disclosure information not submitted.
Kihoon Bohle, BS
Medical Student
Texas A&M School of Engineering Medicine
Disclosure information not submitted.
Currently, management of acute ischemic stroke is dependent on clinical features and radiological biomarkers to determine treatment strategy and prognosis. However, radiological biomarkers must be manually assessed by experts and suffer from observer variability. Additionally, there are a few different methods of mechanical thrombectomy (MT) that can be selected, including thromboaspiration and stent retrieval. While multiple methods may be effective, there is no quantitative system of determining the most effective method to implement. DL and radiomics show promise in (1) guiding optimal MT strategy for a successful recanalization, (2) predicting ischemic stroke outcomes when compared to radiological image biomarkers, and (3) understanding disease characteristics and progression{1}. Recent studies have demonstrated that data-efficient DL models trained to analyze CT angiography data now outperform traditional manually scored radiological image biomarkers in measuring both reperfusion and measuring functional outcomes{2}. In addition, when combined with clinical variables, outcome prediction further improved{3}. AI has potential to improve patient care in acute ischemic stroke management through multiple facets: in diagnosis and characterization of the lesion, guiding treatment through thrombectomy procedures, and predicting future outcomes when compared to standard of care.
Conclusion and/or Teaching Points: