Cerebral arteriovenous malformations (cAVMs) are vascular anomalies in the brain featuring abnormal connections between arteries and veins, and complex anatomy and blood flow. Endovascular embolization is a minimally invasive treatment to occlude cAVMs by injecting liquid embolic agents (LEAs; glue-like substance). However, current embolization strategies often result in incomplete occlusion and may lead to complications such as recanalization (i.e. reopening of occluded vessel). Interventional radiologists currently struggle to predict which part of a cAVM (i.e. which compartment) will be affected by LEA injections and what the impact is on the surrounding vessels. This project aims to develop a novel computational fluid dynamics (CFD) framework to improve the understanding of cAVM hemodynamics and optimize embolization strategies. The project will develop compartment identification algorithms to identify cAVM compartments. Additionally, 3D CFD will simulate the blood flow and LEA distribution in patient-specific cAVM vasculatures and determine the impact of clinically variable parameters (e.g. catheter location, injection timing, etc.). These methods will be validated through a combination of in vivo imaging and in vitro methods. By integrating patient-specific models, advanced material models for LEAs, and compartment identification, this research will enable more accurate predictions of embolization outcomes and support personalized treatment planning towards the future.