Abstract Body: Introduction: Braided flow diverters (FDs) are in widespread use for the treatment of intracranial aneurysms. With iterative FD design improvements, concerns have arisen recently regarding the stability of newer generation FD braids over time. We used the unique prospective Innovative Neurovascular Product Surveillance Registry (INSPIRE, NCT02988128) to compare braid deformations (BDs) and the associated clinical events of an older and newer FD generation (Shield and Vantage, Medtronic, USA).

Hypotheses: a) There is a higher rate of BDs in new vs. older generation FDs, b) BDs are associated with clinical events, c) There are no biomechanical variables associated with BDs, d) BDs can be predicted by artificial intelligence.

Methods: We enrolled 947 patients (524 Shield treated, 423 Vantage) between 2015 to 2020. Clinical events were adjudicated by a clinical events committee. Images were analyzed by two independent readers in a central imaging core lab using recently published criteria. For a radiomics analysis, aneurysm, FDs, and parent artery are segmented. Multiple features are extracted (e.g. tortuosity, diameters along centerlines) as well as deconvolution-based blood flow analysis. Statistical comparisons will be presented. All features are used to train artificial intelligence (AI) models for the occurrence of BD and aneurysm occlusion, for each FD type.

Results: The BD rate was higher in the Vantage cohort compared to the Shield cohort (9.7% vs. 4.5%, p-value 0.002). The analysis of specific patient subgroups that might be prone to BDs is ongoing (e.g. age, sex, co-morbidities). The overall stroke rate was 6.4% (27/423) in the Vantage cohort compared to 4.2% (22/530) in the Shield cohort (n.s.). Of the patients with BD, 1/41 in the Vantage cohort and 2/24 in the Shield cohort had a stroke (2.4% vs. 8.3%, n.s.). The biomechanical variables associated with events and BDs are currently being analyzed and then used to train dedicated AI models.

Conclusions: In this unique comparison, there was a higher rate of BDs in new vs. older generation FDs but no signal that BDs are associated with clinical events. Potential subgroups with higher risk for complications or BDS could help target treatment in the Future. Based on this large, independently adjudicated data base, AI models combining clinical and biomechanical variables will be presented to help optimizing aneurysm patient care.