Automated Segmentation and DIscrimination for Intracranial Aneurysms (AI4IA): Enhancing Robustness of Clinical Risk Scores based on Morphology and Hemodynamics

• Applicants: Privatdozent Dr.-Ing. Philipp Berg, Ph.D.; Professor Dr.-Ing. Leonid Goubergrits
• Subject Area: Fluid Mechanics
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 548907942

Ruptured intracranial aneurysms (IAs) are the main cause of subarachnoid hemorrhage, which can lead to severe disabilities or death. The treatment of unruptured IAs, which occur comparatively frequently, presents a challenge due to numerous uncertainties regarding individual prognosis, optimal therapeutic strategy, and potential treatment complications. Morphological and hemodynamic parameters can contribute to risk stratification but are clinically underutilized due to their lack of robustness. The main objective of the AI4IA project is to identify and to reduce sources of uncertainty in determining morphological and hemodynamic rupture risk parameters to facilitate their translation into patient care. To achieve this, the following main tasks are planned: ● Investigation of possible differences in image data collection between two clinical sites and the creation of a database of real IA geometries. ● Development of standard operating procedures for image data and geometry processing, blood flow simulation, and the determination of morphological and hemodynamic parameters. ● Identification of suitable open-source tools for individual steps to enable easy reproduction of the workflow by third parties. ● Automation of geometry segmentation using artificial neural networks and the subsequent analysis of morphological and hemodynamic parameters. ● Evaluation of the uncertainty of morphological and hemodynamic parameters, computed with the help of machine learning solutions, compared to manual segmentation. ● Identification of an optimal set of robust morphological and hemodynamic parameters to distinguish unruptured from ruptured IAs. Additionally, identification of common characteristics that define robust parameters. The project is designed as a first step towards developing robust rupture risk prediction systems and is designed to last 36 months in the frame of the second SPP2311 funding phase. It is planned as a collaborative project between research groups specializing in neurovascular modeling in Berlin and Magdeburg. Each research group is supported by a clinical partner site. The research groups have extensive experience in modeling IAs, while the clinical partner sites provide the necessary clinical expertise and supply geometries of real IAs in pseudonymized form.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2311: Robust coupling of continuum-biomechanical in silico models to establish active biological system models for later use in clinical applications - Co-design of modeling, numerics and usability
• Co-Investigators: Professor Dr. Daniel Behme; Dr.-Ing. Florian Hellmeier; Professorin Dr.-Ing. Sylvia Saalfeld; Dr. Andreas Spuler