Virtual reality tool corrects errors in brain scan data


Dominique Duncan, assistant professor of neurology at Keck, uses HTC Vive. Surrounding her are MRI images. (Photo courtesy of Dominique Duncan)

When MRI scans are taken, technology automatically splits the data into different sections, often resulting in errors that must be fixed manually. The process of correcting these errors is tedious and time consuming, in addition to teaching students how to do so.

The Keck School of Medicine has developed a solution to this problem — a new technology called Virtual Brain Segmenter.

VBS uses virtual reality features to decipher errors in brain scan data more effectively, instead of relying on computer programs to complete the same task.

“People who had never interacted with MRI data were really excited about the tool,” said Dominique Duncan, an assistant professor of neurology at Keck. “And many commented that it already felt like a game and that they’d love to spend their free time using the tool.”

The idea for this technology was conceived when Duncan was working in one of the neuroimaging labs teaching student volunteers to correct errors in MRI data.

She noticed that students did not like the work, so she decided to create a solution using technology from electronic company HTC’s Vive, a virtual reality headset.

“I was excited when we reached a point where it actually worked, because developing it took some time,” Duncan said. “I loved seeing people’s reactions when they tried it and enjoyed using it.”

Duncan explained that the new technology allows users to zoom in and out of data with ease, while making it easier for them to concentrate on a set of data. A small-size experiment revealed that using the VBS technology saved users 68 seconds when finding data errors in MRI scans, according to USC News.

“It just brings you closer to the data, helping you to better understand its spatial relationships and to provide an intuitive understanding of what goes with what,” said Arthur Toga, director of USC’s Neuroimaging and Informatics Institute.

Now, Duncan plans to work with three L.A. high schools to continue testing VBS before crowdsourcing.

“There are many things that will continue to improve the process, the interface [and] the presentation,” Toga said.

Duncan hopes that VBS will become so intuitive and user-friendly that non-experts can assist with correcting errors.

“As soon as we implement a few extra features, we plan to focus on crowdsourcing this tool and [have] it used by a wide audience around the world,” Duncan said.