RESUMO
Background: Currently, endovascular treatment of intracranial aneurysms (ICAs) is limited by low complete occlusion rates. The advent of novel endovascular technology has expanded the applicability of endovascular therapy; however, the superiority of novel embolic devices over the traditional Guglielmi detachable coils (GDCs) is still debated. We performed a systematic review of literature that reported Raymond-Roy occlusion classification (RROC) rates of modern endovascular devices to determine their immediate and follow-up occlusion effectiveness for the treatment of unruptured saccular ICAs. Methods: A search was conducted using electronic databases (PUBMED, Cochrane, ClinicalTrials.gov, Web of Science). We retrieved studies published between 2000-2022 reporting immediate and follow-up RROC rates of subjects treated with different endovascular ICA therapies. We extracted demographic information of the treated patients and their reported angiographic RROC rates. Results: A total of 80 studies from 15 countries were included for data extraction. RROC rates determined from angiogram were obtained for 21,331 patients (72.5% females, pooled mean age: 58.2 (95% CI: 56.8-59.6), harboring 22,791 aneurysms. The most frequent aneurysm locations were the internal carotid artery (46.4%, 95% CI: 41.9%-50.9%), the anterior communicating artery (26.4%, 95% CI: 22.5%-30.8%), the middle cerebral artery (24.5%, 95% CI:19.2%-30.8%) and the basilar tip (14.4%, 95% CI:11.3%-18.3%). The complete occlusion probability (RROC-I) was analyzed for GDCs, the Woven EndoBridge (WEB), and flow diverters. The RROC-I rate was the highest in balloon-assisted coiling (73.9%, 95% CI: 65.0%-81.2%) and the lowest in the WEB (27.8%, 95% CI:13.2%-49.2%). The follow-up RROC-I probability was homogenous in all analyzed devices. Conclusions: We observed that the coil-based endovascular therapy provides acceptable rates of complete occlusion, and these rates are improved in balloon-assisted coils. Out of the analyzed devices, the WEB exhibited the shortest time to achieve >90% probability of follow-up complete occlusion (~18 months). Overall, the GDCs remain the gold standard for endovascular treatment of unruptured saccular aneurysms.
RESUMO
Gradient-based meta-learning methods have primarily been applied to classical machine learning tasks such as image classification. Recently, PDE-solving deep learning methods, such as neural operators, are starting to make an important impact on learning and predicting the response of a complex physical system directly from observational data. Taking the material modeling problems for example, the neural operator approach learns a surrogate mapping from the loading field to the corresponding material response field, which can be seen as learning the solution operator of a hidden PDE. The microstructure and mechanical parameters of each material specimen correspond to the (possibly heterogeneous) parameter field in this hidden PDE. Due to the limitation on experimental measurement techniques, the data acquisition for each material specimen is commonly challenging and costly. This fact calls for the utilization and transfer of existing knowledge to new and unseen material specimens, which corresponds to sampling efficient learning of the solution operator of a hidden PDE with a different parameter field. Herein, we propose a novel meta-learning approach for neural operators, which can be seen as transferring the knowledge of solution operators between governing (unknown) PDEs with varying parameter fields. Our approach is a provably universal solution operator for multiple PDE solving tasks, with a key theoretical observation that underlying parameter fields can be captured in the first layer of neural operator models, in contrast to typical final-layer transfer in existing meta-learning methods. As applications, we demonstrate the efficacy of our proposed approach on PDE-based datasets and a real-world material modeling problem, illustrating that our method can handle complex and nonlinear physical response learning tasks while greatly improving the sampling efficiency in unseen tasks.