Multimodal deep learning for predicting unsuccessful recanalization in refractory large vessel occlusion

Purpose: This study explores a multi-modal deep learning approach that integrates pre-intervention neuroimaging and clinical data to predict endovascular therapy (EVT) outcomes in acute ischemic stroke patients. To this end, consecutive stroke patients undergoing EVT were included in the study, incl...

Descripción completa

Detalles Bibliográficos
Autores: González Riveros, Jesus David|||0000-0002-9828-0594, Canals, Pere|||0000-0003-0192-3272, Rodrigo-Gisbert, Marc|||0000-0002-7953-3795, Mayol, Jordi, García-Tornel, Álvaro|||0000-0003-3633-3002, Ribó, Marc
Tipo de recurso: artículo
Fecha de publicación:2025
País:España
Institución:Universitat Autònoma de Barcelona
Repositorio:Dipòsit Digital de Documents de la UAB
Idioma:inglés
OAI Identifier:oai:ddd.uab.cat:313046
Acceso en línea:https://ddd.uab.cat/record/313046
https://dx.doi.org/urn:doi:10.1016/j.ejrad.2025.112254
Access Level:acceso abierto
Palabra clave:Acute ischemic stroke
Intracranial atherosclerosis disease
Artificial intelligence
Multimodal deep learning
Explainable AI
Descripción
Sumario:Purpose: This study explores a multi-modal deep learning approach that integrates pre-intervention neuroimaging and clinical data to predict endovascular therapy (EVT) outcomes in acute ischemic stroke patients. To this end, consecutive stroke patients undergoing EVT were included in the study, including patients with suspected Intracranial Atherosclerosis-related Large Vessel Occlusion ICAD-LVO and other refractory occlusions. Methods: A retrospective, single-center cohort of patients with anterior circulation LVO who underwent EVT between 2017-2023 was analyzed. Refractory LVO (rLVO) defined class, comprised patients who presented any of the following: final angiographic stenosis > 50 %, unsuccessful recanalization (eTICI 0-2a) or required rescue treatments (angioplasty +/- stenting). Neuroimaging data included non-contrast CT and CTA volumes, automated vascular segmentation, and CT perfusion parameters. Clinical data included demographics, comorbidities and stroke severity. Imaging features were encoded using convolutional neural networks and fused with clinical data using a DAFT module. Data were split 80 % for training (with four-fold cross-validation) and 20 % for testing. Explainability methods were used to analyze the contribution of clinical variables and regions of interest in the images. Results: The final sample comprised 599 patients; 481 for training the model (77, 16.0 % rLVO), and 118 for testing (16, 13.6 % rLVO). The best model predicting rLVO using just imaging achieved an AUC of 0.53 ± 0.02 and F1 of 0.19 ± 0.05 while the proposed multimodal model achieved an AUC of 0.70 ± 0.02 and F1 of 0.39 ± 0.02 in testing. Conclusion: Combining vascular segmentation, clinical variables, and imaging data improved prediction performance over single-source models. This approach offers an early alert to procedural complexity, potentially guiding more tailored, timely intervention strategies in the EVT workflow.