Graduate Student Profile: Trent Bradberry
Combining Engineering and Neuroscience to Create Assistive Technology
- Hometown: Anderson, S.C.
- B.S.: Computer Engineering, Clemson University, Clemson, S.C.
- M.S.E.C.E.: Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, Ga.
- Advisor: Associate Professor José Contreras-Vidal (Kinesiology, Neuroscience and Cognitive Science Program, Graduate Program in Bioengineering.)
- Started Program: Fall 2004
Completed Program: Spring 2010
Before he came to the University of Maryland, bioengineering student Trent Bradberry was designing signal processing software for Raytheon, a world leader in defense and aerospace systems. With a newly acquired appreciation for signal processing, a background in electrical and computer engineering, and an interest in biology, he began thinking about a new direction: "After about 3 years in industry, I decided that I liked the academic atmosphere and the ability to create and come up with my own ideas," he explains. "I wanted to find a way to combine robotics, biology, and signal processing."
His decision to study the possibility of the human brain's direct interaction with computers wasn't just about a new interest or a degree; it was a decision to change his life, leaving industry for a career in academia and research, exploring his own ideas. He found his opportunity in Associate Professor José Contreras-Vidal's Movement Disorders Lab, part of the Cognitive Motor Neuroscience Lab at the University of Maryland College Park School of Public Health. Bradberry's research is in brain-computer interfaces (BCI), specifically those that are non-invasive.
Devices that make use of the brain's direct interaction with a computer—often known as neuroprosthetics—have the potential to make life easier for people with limited or lost mobility due to injury or disease. For example, a neuroprosthetic could take the form of a robotic arm that reaches, picks up objects, or manipulates other devices for its user. Bradberry's research focuses primarily on helping those with spinal injuries, where the brain is healthy but the body cannot respond to its commands.
In order to understand how the brain controls the body without physically touching the brain, Bradberry uses electroencephalography (EEG), a process that measures electrical activity in the brain by means of a sensor-studded cap that fits tightly on a person's head. Readings are taken externally: no implants are required and there is no physical risk to the subject. Currently, EEG finds most of its clinical use in monitoring and analyzing seizures, comas, diseases of the brain, anesthesia, and sleep. However, Bradberry and researchers with shared interests seek to associate EEG signals with specific tasks, such as reaching for an object or drawing a picture of a specific shape. Bradberry is working on a computer algorithm (a list of instructions for performing certain tasks and recording procedures) that will process a human subject's EEG, understand what is being "asked for", and, ultimately, send those instructions to a neuroprosthesis.
The first part of the work involves developing and fine-tuning the algorithm. Bradberry and the members of the Movement Disorders Lab are studying data from a similar, previous study that used magnetoencephalography (MEG), another non-invasive monitoring technique that records magnetic fields instead of electrical activity. MEG systems have more sensors and record more data, but are not portable, like EEG equipment. So far, the results of the MEG study are promising: Bradberry has been able to decode, in real-time, the trajectory of a person's hand. His hypothesis is that the algorithm developed from the MEG study will be applicable to EEG data they collect. Being able to apply what he has learned from the MEG study to an EEG system that can travel easily with a patient would bring neuroprosthetics closer to becoming a practical reality for people in need.
The second part of Bradberry's work includes recording the arm motion of human subjects reaching in a three-dimensional space and the corresponding EEG data to which the algorithm will be applied. To accomplish this, Bradberry has constructed an apparatus outfitted with buttons and lights that the subjects will be asked to reach for, and the circuitry that interfaces with the EEG computer and a 3-D motion capture system. After all of the data are collected, Bradberry will develop an algorithm that will map the EEG data to the motion data based on the algorithm from the MEG study. This, he says, is the most difficult part of the research. Once the algorithm is complete, a subsequent study will ask subjects to think about moving their arms without actually doing so, and observing whether the algorithm correctly decodes their brain activity and translates it into movements of a robotic arm or a cursor on a computer screen.
How consistent the data will be, how much a limited mobility subject will need to adapt, and how much the algorithm will need to be adapted for any individual remains to be seen. Some flexibility will always be required. One algorithm will not work for all subjects because brain activity varies not only between one person and the next, but also within one person over a period of time. In the future, he and other BCI researchers hope to create interfaces that will automatically support ongoing co-adaptation between user and algorithm.
The Graduate Program in Bioengineering's cross-college, interdisciplinary nature is what attracted Bradberry to the University of Maryland. Without it, he points out, he would not have had the opportunity to seek out and work with Dr. Contreras-Vidal, whose home department, Kinesiology, is not part of the A. James Clark School of Engineering. In fact, it was because of Bradberry's interest in having him as an advisor that Contreras-Vidal became involved in the program.
Being able to establish that relationship has had a greater impact than simply finding the right lab. "The best thing about my experience here has been my advisor allowing me to come up with my own ideas and actually try them," he says. "He's not just saying 'Go do this and send me the results.' He's allowing me to say, 'Well, what if we try this or do this instead?' He's been very supportive, and it's had a huge impact on me."
After graduating, Bradberry would like to pursue an academic career, and continue to further brain-computer interface technology. "I don't see myself founding a company to do this," he says, "but I do forsee a time when this technology will be mature enough to transfer to someone [in industry] who could get it to the people who need it. I would want to continue furthering the research." He is enthusiastic about becoming a professor. In the spring of 2007, he was one of only 20 students chosen to participate in the Clark School's new Future Faculty Program, which was created to prepare graduate students for academic careers in top-50 engineering schools. The program includes seminars, a teaching practicum, and a research mentoring practicum.
Bradberry recommends that any undergraduates interested in applying what they know to biology or medicine consider a graduate degree in bioengineering. What's not always clear to people, he says, is where and how their current interests or backgrounds could fit into a bioengineering program. He suggests getting started by talking to students and faculty at schools of interest and getting involved with research that integrates one's current skills with the life sciences.
Since unlike many students he spent some time in industry before deciding to pursue a Ph.D. and an academic career, Bradberry doesn't rule out trying other job sectors either: "After my master's degree, I was really set on industry and really thought that was where I wanted to be long-term, but changed my mind. I wouldn't make a blanket statement that someone should go into one sector over another—it's a personal choice. But I would certainly encourage people to be open to change."
Learn more about the promising results of Trent's work:
See "Reconstructing Three-Dimensional Hand Movements from Noninvasive Electroencephalographic Signals," Trent J. Bradberry, Rodolphe J. Gentili, and José L. Contreras-Vidal. The Journal of Neuroscience, 30(9):3432-3437 »