Artery Visualization Gets to the Heart of the Matter

A team of Harvard researchers has developed a new arterial visualization tool that may result in quicker and more accurate diagnoses of heart disease.

The interactive visualization tool, called HemoVis, takes data from patient-specific blood flow simulations and flattens out the arterial geometry into a 2-D tree diagram representation, which highlights at-risk areas in a single red-to-black color scheme. The method departs from traditional artery visualization tools, which require users to rotate complex 3-D models that use rainbow color mapping.

“Our goal was to develop the most accurate and efficient visualization of arteries for patient diagnosis,” said Michelle A. Borkin ’06, a Harvard doctoral candidate in applied physics and lead author of the project.

The research was mainly a collaboration among the School of Engineering and Applied Sciences, Harvard Medical School, and Brigham & Women’s Hospital. Working with multiple physicians and researchers, the team compared the effectiveness of both techniques in identifying regions of low endothelial shear stress within arteries, an occurrence linked to the progression of disease in coronary arteries. By allowing researchers to simply observe 2-D cross sections rather than spinning around 3-D images, the HemoVis tool resulted in fewer mistakes in identification.

“Even experts missed a lot of the at-risk areas with the traditional 3-D models,” said Amanda E. Peters, team researcher and a Harvard doctoral candidate in applied physics.

Through multiple clinical simulations, the HemoVis tool was shown to significantly increase diagnostic accuracy from 39 percent to 91 percent.

The overall simulation project has spanned nearly five years. The first iteration of the tool actually emerged from Borkin’s final project in Computer Science 171: “Visualization”, taught by her current graduate advisor Professor Hanspeter Pfister.

Though the team did acknowledge the continued importance of 3-D color mapping tools in surgical planning, the findings met resistance within the medical visualization community in their repudiation of the long-held belief that color mapping techniques were superior.

“People were so set on how they were used to seeing the data,” Peters said. “It took a lot of effort to convince people to accept a different perspective.”

The tool gained wide acceptance during its presentation Saturday at the IEEE Information Visualization Week, one of the leading visualization conferences held this past weekend at Brown University.

The developers are now working to add new functionality to optimize its effectiveness.

“In our current and future work, we are continuing to improve the 2-D interface as well as link the 2-D visualization to its 3-D representation to assist in other clinical tasks such as surgical planning,” Borkin said.

The team’s findings will be published later in the year in the journal IEEE Transactions on Visualization and Computer Graphics.

—Staff writer Akua F. Abu can be reached at aabu@college.harvard.edu.