Smarter flow simulation model for analyzing blood flow in brain aneurysms

To simulate blood flow inside brain aneurysms, researchers from Japan developed a computational method that combines 4D flow MRI, computational fluid dynamics, and data assimilation, which provides greater accuracy and efficiency. By focusing only on the aneurysm region, this approach significantly reduces computational cost while improving flow estimation. When validated on patient data, it outperforms conventional models—offering a practical tool for patient-specific risk assessment and treatment strategies.

Brain aneurysms, also known as cerebral aneurysms, are pathological dilations of blood vessel walls that form bulges in blood vessels of the brain. These bulges form due to weakening of the blood vessels and can arise from different conditions like high blood pressure, genetics, unhealthy lifestyle, or underlying heart conditions. Rupture of brain aneurysms is a serious event that can lead to stroke or even death. Therefore, to assess the risk of rupture, it is important to understand how blood flows inside a brain aneurysm.

Many blood flow simulation methods exist to date, including those using phase-contrast magnetic resonance imaging (MRI), also known as four-dimensional (4D) flow MRI, and computational fluid dynamics (CFD)—a computational simulation for the flow of fluids. But these methods often require higher spatiotemporal resolution or lack patient-specific data.

Another approach is optimizing the methods with variational data assimilation (DA)—a technique that combines observational data with a numerical model. However, the models developed with this technique often require high computational costs arising from analyzing the entire main vessels, in addition to the aneurysms, limiting their practical use in clinical settings.

In this context, research team led by Professor Satoshi Ii from the Department of Mechanical Engineering, School of Engineering, Institute of Science Tokyo (Science Tokyo), Japan, have developed a practical and efficient strategy to estimate blood flow within brain aneurysms using a smarter combination of 4D flow MRI and CFD with DA. The findings were made available online on May 19, 2025, and will be published in Volume 268 of the journal Computer Methods and Programs in Biomedicine on August 01, 2025.

Unlike previous models that require full vessel data, this new approach focuses only on the aneurysm region, significantly reducing the computational costs. Moreover, it uses limited MRI data to specifically analyze the flow velocities near the aneurysm neck, which is required for input in CFD, and it uses the variational DA method to estimate the full velocity profile inside the aneurysm.

"Our method avoids modeling the entire vascular system," explains Ii, "Even with minimal data, we could achieve simulations that match patient-specific blood flow patterns with remarkable accuracy."

The researchers validated their method using both synthetic data and real patient datasets. When tested with simulated data, the velocity mismatch between the developed model and ground truth was only 4%–7%. Whereas in tests with MRI data from three patients, the method reduced its velocity errors by 37%–44% in comparison to traditional 4D flow MRI and CFD models and was therefore more efficient. In effect, it captured more realistic flow patterns using limited data and minimum computing power.

The key innovation of the method was the use of “Fourier series-based model order reduction,” a mathematical optimization technique that simplified how the time-varying flow of blood was represented. This significantly reduced the computational complexity and avoided errors in fluid dynamics. Additionally, not only did the new model simulate flow with better accuracy, but it also offered clearer insights into critical hemodynamic factors such as wall shear stress and pressure of the flow.

By harnessing the data assimilation to focus on the aneurysm zone, the method bypasses the challenges of extracting clean boundary conditions from noisy MRI scans of full vessel branches, making the method more efficient and robust for clinical applications.

“This approach could become a valuable tool for neurosurgeons and radiologists," concludes Ii, “The quantitative evaluation of patient-specific blood flow using this method may aid future predictions of aneurysm growth and rupture, supporting early medical decisions and better management.”

This work was supported by the MEXT Program for Promoting Research on the Supercomputer Fugaku (Development of human digital twins for cerebral circulation using Fugaku, JPMXP1020230118) and JSPS KAKENHI Grant Number JP22H00190, JP24K02557. Part of research was conducted using computational resources of the supercomputer Fugaku provided by the RIKEN Center for Computational Science (project ID: hp230208, hp240220, hp250236) and the supercomputer "Flow" at the Information Technology Center, Nagoya University.

***

About Institute of Science Tokyo (Science Tokyo)
Institute of Science Tokyo (Science Tokyo) was established on October 1, 2024, following the merger between Tokyo Medical and Dental University (TMDU) and Tokyo Institute of Technology (Tokyo Tech), with the mission of “Advancing science and human wellbeing to create value for and with society.”