Hearing a series of quick scratching sounds is not generally cause for celebration. But for Leigh Hochberg the sounds represented a scientific breakthrough that could enable paralyzed people to interact with the world more easily.
Hochberg, a professor of engineering and co-director of the BrainGate2 research effort, first heard the scratching sounds, which represent neural impulses, in 2005. Almost seven years later, researchers at Brown are forging ahead and collaborating with new partners. Early last month, Stanford came on board, and researchers at the university are actively enrolling participants for clinical trials of a system that would allow paralyzed people to manipulate machinery with their minds.
Turning thought into action
Most people take for granted the complicated series of biological signals that translate a thought into action. The process that allows us to think about grabbing a glass of water in front of us and almost instantly move our hand can be harnessed to allow quadriplegic people to control external devices.
The clinical trials test three main aspects of the BrainGate2 neural interface, said John Donoghue, professor of neuroscience, director of the Brown Institute of Brain Science and Hochberg’s co-leader on BrainGate2, a continuation of the original BrainGate project.
The first step in translating neural signals into the manipulation of a device is a sensor, a small square the size of a baby aspirin with 100 thin, millimeter-long spikes. The sensor picks up and records the firing patterns of anywhere from a few to 150 of the brain’s 86 billion neurons. The device is implanted into the area of the motor cortex that controls arm and hand movement, Donoghue said, because the paralyzed trial participants control objects like cursors that are generally operated with those body parts.
Training computers to work with brains
Michael Black, professor of computer science and a BrainGate2 researcher, said his work focuses on decoding neural signals into patterns that can be understood by machines.
“The work that my group has been doing involves mathematically modeling how the activity of a population of neurons in the brain is related to movement or imagined movement,” he said.
Computers need to be trained to recognize neural patterns and respond to them, Black said. During each trial, the parameters of the computer models that decode neural signals need to be adjusted because the firing rates of neurons vary depending on a variety of factors like the noise level in a room or the subject’s energy level, Black said. The goal is to compile large amounts of “training data” to help the computer eventually adjust the models on its own.
“The whole reason we care about that is to make these things more robust,” Black said. “We want to take this whole research from a very simple level in a controlled environment and make it more practical for people.”
Decoding the neural impulses is necessary so users can activate and operate different devices. The first device that was operated in a clinical trial was a simple mechanical hand that opened and closed. Donoghue said other devices tested have included a cursor, a small robot that grabbed a piece of candy and a primitive wheelchair.
Researchers in Hochberg’s lab are working on new communication devices. Undergraduate researcher Kathryn Tringale has been designing more intuitive keyboards that would be easier for participants to control with their minds than standard QWERTY keyboards. The standard keyboard, designed to prevent typewriter jams, is organized so letters that frequently appear next to each other in writing are not near each other on the keyboard, an impractical design when the cursor is controlled mentally.
Tringale said she has been studying mathematical models of the brain to better understand what a practical communication device would look like and to create a model of a neuron-controlled cursor that she uses to compare the different keyboards she has designed.
Currently, many people with locked-in syndrome, a condition that prevents almost all movement, communicate with letter boards. Participants must go through each letter until they get to the one they want and indicate it through eye movement, painstakingly spelling out words and sentences. Tringale said observing this time-consuming process has motivated her research. “Actually having that interaction and experiencing it in real time really inspires you,” she said.
Researchers at Case Western Reserve are also investigating devices that could greatly impact those living with paralysis. Hochberg said Case Western’s Functional Electrical Stimulation Center is researching technology to restore simple movements in paralyzed people by stimulating the nerves in the arms and legs with implanted electrodes that can be controlled through simple switches. Hochberg said he hopes this technology could one day be combined with BrainGate2 so people’s thoughts could again control their own limbs, rather than an external device.
‘Remarkable participants’
BrainGate2 has enrolled seven participants to date, two of whom are currently enrolled. Hochberg said Brown and Stanford researchers have permission to enroll up to 15 participants.
“Largely, the participants or their families find us,” Hochberg said. Once participants make an initial contact and if they seem to meet the inclusion criteria, Hochberg visits them at their homes to describe the study and answer their questions.
Participating in the trial requires a significant commitment — in addition to attending half-day research sessions twice a week for at least 13 months, participants also must undergo surgery to implant the neural sensor in their brains.
Both Hochberg and Donoghue said interacting with participants is the most important and the most rewarding part of their work.
“We could not develop this technology without the feedback of remarkable participants,” Hochberg said.
Donoghue said the project still has a long way to go.
One of the many challenges now is the development of a wireless implant system. Right now, battery technology is insufficient and neural implants must be plugged into a power source. The challenge is creating a battery whose lifespan is long enough to merit implantation with the knowledge that a participant will be forced to undergo neurosurgery each time the battery runs out. Donoghue said he thinks the team will soon be able to create a battery with a 10-year lifespan.
The process of decoding neural signals is also very complex, and the team is still striving to create the tools that will allow people to control devices with as much ease and fluidity as able-bodied people control their own bodies.
“Working on BrainGate has taught me that we can do basic science and have a profound impact on people’s lives at the same time. … I want everything I do to have an impact like that,” Black said. “It’s changed me profoundly.”