Issue: Volume: 23 Issue: 9 (September 2000)

Secrets of the Heart

By Diana Phillips Mahoney


For years, engineers have been using computer visualizations of simulated flow data to better understand how fluid moves around and through complex mechanical assemblies and how that flow affects and is affected by structural design characteristics. Now, one of the most complex natural assemblies-the human heart-is benefitting from the same visual insight.

Cardiologist Ann Bolger at the University of California at San Francisco, with the help of research colleagues at Sweden's Linkoping University, is using traditional engineering-visualization software to obtain unprecedented perspectives on the movement of blood within the human heart-information that puts into question long-held notions of fundamental heart physiology.

The research team has developed a way to import 3D MRI data from a live, beating heart into a computational fluid dynamics (CFD) visualization program to follow the path of blood through the heart chambers. Using EnSight visualization software from CEI (Morrisville, NC), cardiac blood-flow characteristics can be depicted using conventional streamline and pathline techniques as well as fully animated particle traces. "The very first time we actually were able to get a dataset in [to EnSight], we said, 'Wait a minute. The blood flow doesn't do that.' Then we said, 'Oh, maybe it does do that.' We realized we were seeing things that no one was aware of before," says Bolger.

The human heart, in good working order, epitomizes the expression "a well-oiled machine." It moves blood rhythmically and tirelessly through the labyrinthine cardiac channels, delivering it throughout the body to sustain life. When it works, it really works. When it doesn't, well, that's a problem. The slightest structural, functional, or electrical malfunction can wreak havoc on the vascular system, with potentially devastating consequences. Identifying the source and effect of such irregularities has been hampered by the limitations of traditional imaging techniques, says Bolger. "The holes in our knowledge have to do with the global patterns of flow inside the heart that affect how well the heart can perfuse the body, how well the valves open and shut," she says. "But our knowledge of the interaction of global flow and anatomic structure, such as walls and valves, is not good at all-not because people aren't interested, but because we haven't had the right tools."
CEI's EnSight software allowed researchers to create this particle trace visualization of blood flow through the heart. The artificial particles were traced both backward and forward in time to provide a more complete representation of flow. (Image co

Traditional modes of cardiac-flow investigation fall into two categories: projection techniques, such as X-ray imaging, and ultrasound. The X-ray approach involves injecting a contrast dye into the patient's blood and taking a picture. Unfortunately, says Bolger, "you're looking through a 3D structure and creating a flat image from that. Also, there's no way of tracking individual areas of flow." Ultrasound techniques are considered more useful than projection methods because they are able to provide detailed images of walls and valves, as well as accurate velocity information at specific sites in the heart. The downside, however, is that ultrasound is a planar technique, "so you're only estimating velocity that's in that plane," says Bolger. "If a blood cell flows through the plane obliquely, it's misrepresented. There's no way to use that planar technique-even if you reconstruct it in 3D-to really tell the truth about the paths that blood actually follows in 3D space inside the heart."

It was Bolger's frustration with these deficiencies, as well as her awareness of a general, widespread lack of understanding of basic aspects of cardiac function among physicians, that spurred her quest for a better imaging approach-a journey that ultimately led to magnetic resonance imaging (MRI), a technique that uses a magnetic field and radio waves in conjunction with computer technology to produce high-resolution images of organs and fluid flow. However, obtaining accurate flow measurements near heart valves, even with advanced time-resolved MRI techniques, is challenging. The heart contracts along both short and long axes during the cardiac cycle (one heart beat), and the consequential flow implications are difficult to represent through the use of single acquisition planes or imaging "slices" through the region of interest, which is what MRI provides. This can lead to the misregistration of cardiac anatomy and distortions in blood velocity data.

"MR velocity is always a 3D velocity, but with a single acquisition plane, even if you let the velocities go out of the plane, you don't know where they go from there," says Bolger. "So while we need the temporal resolution, because each step of the cardiac cycle has discrete phases where the chambers fill and empty, we also need the accurate 3D dataset, since flow may start in one point and follow a very curving, complicated route."

In engineering, the best way to follow such complicated pathways is through the development of a computational grid to represent the 3D volume of a flow environment. The cardiac researchers realized they could apply the same paradigm to the heart, and so they did, with the only difference being that they were starting with an actual physiological model rather than a mathematical simulation.

To build the blood-flow model, the researchers acquire MRI velocity data during 32 phases of the cardiac cycle, starting at a specified time and location. "Starting at the same point in the cycle, we take hundreds of beats and put them together to get a representative dataset [of the blood's path through the heart]," says Bolger. The collected data is then corrected for such effects as eddy currents and aliasing, and the format is changed to suit the EnSight environment. From this point, the visualization process is the same as that implemented in engineering applications.

The use of particle paths, in which particles are emitted at a given location and transported by the fluid flow, is particularly well suited to blood flow analyses, says Bolger. "Particles are what cardiovascular flow is about. In this case, the particles are the red cells. So it's a very relevant model for asking the question: How does Red Cell Number 1 get from Point A to Point B compared to Red Cell Number 2, which comes from a different part of the heart?"

Not only is the technique relevant in a physiological sense, the animation capabilities are also conceptually fitting. "When you animate 3D pathlines, the image starts to look much like blood moving through the heart should look," Bolger says. "There's so much information in the images that it is critically important for them to be presented in a form that helps you make sense of data." As in engineering applications, the fluid particles can be colored to represent different characteristics of the flow such as velocity and pressure.
This series of frames from a particle trace animation of blood flow within the heart depicts (beginning at the lower left pulmonary vein) the swirling blood flow that creates the vortex within the heart's atrium.

Ironically, the researchers' greatest challenge in bringing this imaging and visualization approach to fruition has little to do with the visual technologies themselves. "One of the most difficult things is presenting it to our colleagues in a way that helps them let go of what they think they know," says Bolger. "For example, we had the hardest time convincing one of the world's top cardiac ultrasound authorities that the blood flow didn't just come into the heart, stop, then get pushed out the other way, but that it in fact flows in these beautiful and elegant pathways."

Once further developed and accepted as a viable approach, the blood-flow visualization technique could play a significant role in cardiac treatment planning. The first step in this direction is the creation of image databases representing different cardiac states. "The first thing we did was to create velocity visualizations for a population of normal [with respect to cardiac structure and function] people over a wide range of ages, because we needed to know what normal looks like," says Bolger.

This information is then used as the control data for comparative studies. Already it's been compared against the visualizations made from people with high blood pressure, in whom the only noticeable difference when studying traditional ultrasound images of their hearts was a slight thickening of the cardiac walls. With the MRI flow visualization, says Bolger, "there were dramatic differences in the distribution of flow in the filling phases of the heart cycle. What we were able to see that we never could have realized with ultrasound was the location of some of the differences and what kind of impact they might have." Such information could, for example, be invaluable in planning valve repair or replacement surgery, she says.

With its potential to change long-held conceptions about cardiac physiology and pathophysiology, flow visualization could rev olutionize cardiac health care, Bolger says. "The heart is one of the best machines that has ever been created. Being able to study how this pump can create the flows that it does, under the circumstances that it does, is really an opportunity to understand effects that could help significantly. We're looking at basic cardiovascular physiology that hasn't really been studied before."

EnSight, CEI (