Researchers at UCLA are generating flow models of blood through brain aneurysms with the goal of better predicting how the aneurysm will grow and then to more accurately formulate a treatment. CFD results produce revealing flow patterns and calculate shear stresses on vessel walls. Accurate images of brain blood vessels and aneurysms come from X-ray computed tomography (CT), digital angiography, and a more recent technology called 3D rotational digital angiography.

Treatment for aneurysms includes Guglielmi detachable coils or GDCs. Placing a coil, which looks like a small tangle of wire, into an aneurysm encourages blood clotting to seal it off and limit its growth. CFD results also show how the blood velocity changes as it encounters the coils. Without treatment, an aneurysm could swell and break, causing a stroke and death. Previous treatment was surgery, which is more traumatic for patients.

“We can image blood vessels in great detail with x-ray-based techniques,” says Assistant Professor Daniel Valentino, Chief of Imaging Informatics at the University of California, Los Angeles. “We are working on algorithms to extract vascular structures, and create models and meshes of the image data. This preprocessing readies the images for flow simulations which would use either Navier-Stokes approaches or lattice Boltzmann method. The later is more useful for simulating flow fields around complex objects such as GDCs. The method is also easily adjusted to run on computers with many processors.”

“The research here is looking for quantitative information to let us better diagnose and evaluate patients with cerebrovascular disease,” says Valentino. It causes aneurysms and is the third leading cause of death and the leading cause of nursing home admissions in the U.S.

The problems Valentino's team has encountered include preprocessing the model, long run times, and validating results. Most simulation tools don't make it easy to do the whole modeling process, he says, so a setup takes several days. And then solution times run 30 to 40 days on single-processor computers. One way to validate the accuracy of the simulations compares them to in vitro measurements, such as results from laser Doppler veloscimetry or particle-image veloscimetry. But even those have inherent errors.

“Our goal is to develop a combination of tools that streamline going from acquiring medical imaging data to simulation results. We are using commercial tools and in-house developed software to build and streamline the pipeline for processing data. Then we want to know if the simulation results match what we observe in patients, and whether or not treatments proscribe guided by simulations correlate well with the outcomes observed in patients,” he says.