He captivated the Lipsett Amphitheater audience with a 2013 video clip: Outfitted with an EEG cap and a laptop, a graduate student used his thoughts alone to pilot a model helicopter around an obstacle course set up in a gymnasium.
Distinguished McKnight University professor of biomedical engineering, director of the Institute for Engineering in Medicine at the University of Minnesota, and an NIBIB and NEI grantee, He had earlier in his lecture laid the foundation for the video. He outlined the numerous complexities involved in mapping the mind and the capabilities—and limitations—of the field’s latest neuroimaging technologies.
“Our brain activity is not just distributed over the three spatial dimensions,” he explained, “so we need high-spatial resolution imaging. But it also is really a dynamic process…What we really want is to map the information-processing process, not just say where the neuron is located.”
Current clinical practice requires invasive methods in order to localize a seizure as it is occurring. In a surgical procedure, doctors must insert probes or sensors through the skull into specific brain regions.
The top two noninvasive tools, however—functional magnetic resonance imaging (fMRI) and EEG—have their own drawbacks as well. “fMRI is very precise in spatial domain, however it is limited by its slow response in the time domain,” He pointed out. “EEG is very fast in time, but relatively low in spatial resolution.”
What if we could somehow combine measurements from two or more methods? That’s “multimodal neuroimaging” and it’s what He’s group and others have been working on for the past decade or so.
“There’s a lot of work that has been done to try to improve the EEG spatial resolution by solving for something called electrical source imaging—ESI—and integrating EEG with functional MRI,” He said, describing how his group mapped brain dynamics using EEG recordings, then added fMRI data.
A graduate student—outfitted with an EEG cap and a laptop—uses his thoughts alone to pilot a model helicopter around an obstacle course set up in a gymnasium.
Lecture photos: Ernie Branson, flight demo photos: University of Minnesota
|Director of the Institute for Engineering in Medicine at the University of Minnesota, He explains the flight demonstration at a recent NCCAM lecture. He is a grantee of NIBIB and NEI.
Billions of neurons work in the brain, he explained. “To map or measure a single neuronal response, an EEG is definitely not your choice. There is [currently] no noninvasive neuroimaging technology to accomplish that goal.”
He explained that an EEG recording of the dynamic action of neurons delivers what’s known as a “synchronized neural network response.” By recording EEG as a “filter” to responses produced by all the neurons, however, scientists are able to isolate the neural network responses that encode the brain function. With even more sophisticated math, He reported, researchers can now map the 3D whole human brain function using electrical source imaging.
“There is a need for engineering innovation to solve such a [source-imaging] problem,” He admitted, “but the good news is that we have done a lot in the past to accomplish a much more reasonable spatial resolution from the EEG.”
To test a clinical application of the method, He, who also serves as a consultant on NIH’s BRAIN Initiative, examined how pain can be noninvasively quantified and imaged in a thermal stimulation study. Researchers established a temperature, asked subjects to provide a pain rating and then recorded an EEG.
“We tried to see if we could derive quantitative and objective biomarkers to quantify the pain level and try to image the neural network involved with pain,” He said. Results were encouraging.
Noninvasive epilepsy localization presents another potential clinical use for He’s work. About 30 percent of people with epilepsy do not respond to anti-seizure medication. Patients in that group who experience severe seizures often need surgery to remove the seizure-causing region of their brain. To find out precisely where in the brain the seizures occur, doctors perform open-skull monitoring—also a surgical procedure.
Other problems exist with the monitoring, too: seizures can’t be predicted and the procedure is expensive, offers limited cortical coverage and requires a stay in intensive care. What if a non-invasive imaging technology could pinpoint the seizure region just as well? He’s group tested its method in collaboration with physicians at Mayo Clinic; accuracy compared favorably with the surgical monitoring.
He, with the mind-maneuvered model ’copter navigating through rings suspended in a gym
|Before the lecture, NCCAM’s deputy director Dr. David Shurtleff (l) and director Dr. Josephine Briggs welcome He.
He devoted the last third of his lecture to a topic dear to the hearts (and minds) of folks at the National Center for Complementary and Alternative Medicine, whose integrative medicine research series sponsored He’s visit.
“From the EEG, from the neurophysiology and the hemodynamic responses, one thing we can do is to understand the brain,” he noted. “But there is one more step. By understanding the brain, we want to aid diagnosis and treatment of disorders. But can we help the patient accomplish something more? That is really the scientific goal of the brain computer interface field.”
He’s lab used electric source imaging to investigate where in the brain a signal comes from when you imagine a motion. For example, when a person who is paralyzed thinks about moving a limb—but can’t move it—where does that thought originate? By determining what’s called the “event-related desynchronization,” or ERD signal, and where it’s located, scientists can understand how such a cue is generated and how to design a better system.
Mapping and decoding motor imagination eventually led He and colleagues to develop the technology that allows users to fly toy choppers using thoughts only.
In an interesting sidenote, He and colleagues just by chance discovered that people with previous mind-body awareness training—like yoga or meditation—perform better on BCI tasks. A research paper on that finding recently was accepted for publication.
“We learned from this EEG mapping work to optimize our system and to demonstrate for the first time in the world that noninvasive brain computer interface is able to control a real physical device for sophisticated tasks in 3D physical space,” He concluded. “Of course my ideal case is to control a robotic arm helping a patient with prosthetics, but we have to test on healthy subjects first.”
He’s full lecture is available online at http://videocast.nih.gov/summary.asp?Live=14515&bhcp=1.