Periodic Reporting for period 1 - CURE (Fluid-Structure Interaction and Machine Leaning for Controlling Unruptured Intracranial Aneurysms)
Reporting period: 2023-01-01 to 2025-06-30
Intracranial aneurysms affect 3–5% of the population and are responsible for most non-traumatic subarachnoid hemorrhages—with fatality rates reaching 50% and significant long-term disability among survivors. Current treatment options, including surgical clipping, coiling, and flow diverter stents, rely on subjective assessments based on aneurysm size, shape, and risk factors. In this context, the ERC “CURE” aims to transform clinical decision-making by developing a next-generation computational framework that integrates state-of-the-art computational fluid dynamics (CFD), fluid–structure interaction (FSI), and artificial intelligence. By automating image segmentation, adaptive meshing, high-fidelity hemodynamic simulations, and deep reinforcement learning (DRL) for patient-specific stent optimization, we strive to provide objective, quantitative tools that not only predict aneurysm rupture risk but also guide personalized treatment planning. The ultimate goal is to support clinicians with accurate, rapid, and reliable simulation outputs that can save lives and reduce the societal burden of suboptimal aneurysm management.
Our team has developed a cutting-edge computational framework that is revolutionizing how aneurysms are treated. One of the main achievements is the amazing ability to create patient-specific stents using advanced AI and modeling. This technology tailors the stent design to each patient's unique vascular geometry, adjusting key features like wire count, porosity, and braiding angle. This optimization has proven to be a unique capability, offering personalized solutions that improve patient treatment outcomes. By using deep reinforcement learning (DRL), we are able to calculate precisely the ideal stent configuration every time, based on a patient's anatomy and blood flow dynamics. This has allowed us to significantly enhance the treatment of aneurysms, saving lives and offering real-time, accurate solutions for critical decisions. CURE have already assisted a patient using its high fidelity simulation aided clinicians to confirm that the patient’s stent placement was effectively occluding the aneurysm by 58%, and predicted that a second stent might be necessary—a major step in life-saving intervention. Alongside the stent design progress, we have worked on Fluid-Structure Interaction (FSI), a key element of our work that models how blood flow and vessel walls interact. By simulating this dynamic relationship, we can create better, more realistic representations of what happens inside the human body. Thanks to this innovative combination of FSI and AI, we now have predictive methods that can guide medical professionals' decision-making, offering more accuracy and reducing uncertainty during procedures. These achievements stand out because they enable highly precise, personalized approaches to treating aneurysms, proving vital in time-sensitive, life-or-death decisions. This work not only pushes the boundaries of computational sciences but is directly helping clinicians improve patient care.
Our work represents a major leap forward in aneurysm management by introducing three groundbreaking innovations. First, we have created a unique, large-scale dataset for fluid–structure interaction (FSI) that captures the detailed interplay between blood flow and vessel wall dynamics—an unprecedented resource that significantly enhances our understanding of aneurysm risk factors. Second, we have developed a truly patient-specific stent design that is tailored to the unique vascular anatomy of each individual. This innovative approach combinng machine learning and computational mechanics departs from the traditional stent model, ensuring optimal stent performance and improved treatment precision. Finally, our high-fidelity simulation framework enables accurate risk assessment by realistically modeling both blood flow and vessel, thereby providing clinicians with actionable insights into rupture risk and treatment outcomes. Collectively, these advances set new benchmarks in personalized medicine and data-driven healthcare, paving the way for rapid clinical translation and improved patient care.