Brown unveils novel wireless brain sensor

In a significant advance for brain-machine interfaces, engineers at Brown University have developed a novel wireless, broadband, rechargeable, fully implantable brain sensor that has performed well in animal models for more than a year. They describe the result in the Journal of Neural Engineering and at a conference this week.

PROVIDENCE, R.I. [Brown University] — A team of neuroengineers based at Brown University has developed a fully implantable and rechargeable wireless brain sensor capable of relaying real-time broadband signals from up to 100 neurons in freely moving subjects. Several copies of the novel low-power device, described in the Journal of Neural Engineering, have been performing well in animal models for more than year, a first in the brain-computer interface field. Brain-computer interfaces could help people with severe paralysis control devices with their thoughts.

Cortex communication
Engineers Arto Nurmikko and Ming Yin examine their
prototype wireless, broadband neural sensing device.
Credit: Fred Field for Brown University

Arto Nurmikko, professor of engineering at Brown University who oversaw the device’s invention, is presenting it this week at the 2013 International Workshop on Clinical Brain-Machine Interface Systems in Houston.

“This has features that are somewhat akin to a cell phone, except the conversation that is being sent out is the brain talking wirelessly,” Nurmikko said.

Neuroscientists can use such a device to observe, record, and analyze the signals emitted by scores of neurons in particular parts of the animal model’s brain.

Meanwhile, wired systems using similar implantable sensing electrodes are being investigated in brain-computer interface research to assess the feasibility of people with severe paralysis moving assistive devices like robotic arms or computer cursors by thinking about moving their arms and hands.

This wireless system addresses a major need for the next step in providing a practical brain-computer interface,” said neuroscientist John Donoghue, the Wriston Professor of Neuroscience at Brown University and director of the Brown Institute for Brain Science.

Tightly packed technology

David Borton
"The first fully implanted microsystem operated
wirelessly for more than 12 months in large animal
models - a milestone."

In the device, a pill-sized chip of electrodes implanted on the cortex sends signals through uniquely designed electrical connections into the device’s laser-welded, hermetically sealed titanium “can.” The can measures 2.2 inches (56 mm) long, 1.65 inches (42 mm) wide, and 0.35 inches (9 mm) thick. That small volume houses an entire signal processing system: a lithium ion battery, ultralow-power integrated circuits designed at Brown for signal processing and conversion, wireless radio and infrared transmitters, and a copper coil for recharging — a “brain radio.” All the wireless and charging signals pass through an electromagnetically transparent sapphire window.

In all, the device looks like a miniature sardine can with a porthole.

But what the team has packed inside makes it a major advance among brain-machine interfaces, said lead author David Borton, a former Brown graduate student and postdoctoral research associate who is now at Ecole Polytechnique Federale Lausanne in Switzerland.

“What makes the achievement discussed in this paper unique is how it integrated many individual innovations into a complete system with potential for neuroscientific gain greater than the sum of its parts,” Borton said. “Most importantly, we show the first fully implanted microsystem operated wirelessly for more than 12 months in large animal models — a milestone for potential [human] clinical translation.”

The device transmits data at 24 Mbps via 3.2 and 3.8 Ghz microwave frequencies to an external receiver. After a two-hour charge, delivered wirelessly through the scalp via induction, it can operate for more than six hours.

“The device uses less than 100 milliwatts of power, a key figure of merit,” Nurmikko said.

Co-author Ming Yin, a Brown postdoctoral scholar and electrical engineer, said one of the major challenges that the team overcame in building the device was optimizing its performance given the requirements that the implant device be small, low-power and leak-proof, potentially for decades.

“We tried to make the best tradeoff between the critical specifications of the device, such as power consumption, noise performance, wireless bandwidth and operational range,” Yin said. “Another major challenge we encountered was to integrate and assemble all the electronics of the device into a miniaturized package that provides long-term hermeticity (water-proofing) and biocompatibility as well as transparency to the wireless data, power, and on-off switch signals.”

With early contributions by electrical engineer William Patterson at Brown, Yin helped to design the custom chips for converting neural signals into digital data. The conversion has to be done within the device, because brain signals are not produced in the ones and zeros of computer data.

Ample applications

The team worked closely with neurosurgeons to implant the device in three pigs and three rhesus macaque monkeys. The research in these six animals has been helping scientists better observe complex neural signals for as long as 16 months so far. In the new paper, the team shows some of the rich neural signals they have been able to record in the lab. Ultimately this could translate to significant advances that can also inform human neuroscience.

Current wired systems constrain the actions of research subjects, Nurmikko said. The value of wireless transmission is that it frees subjects to move however they intend, allowing them to produce a wider variety of more realistic behaviors. If neuroscientists want to observe the brain signals produced during some running or foraging behaviors, for instance, they can’t use a cabled sensor to study how neural circuits would form those plans for action and execution or strategize in decision making.

In the experiments in the new paper, the device is connected to one array of 100 cortical electrodes, the microscale individual neural listening posts, but the new device design allows for multiple arrays to be connected, Nurmikko said. That would allow scientists to observe ensembles of neurons in multiple related areas of a brain network.

The new wireless device is not approved for use in humans and is not used in clinical trials of brain-computer interfaces. It was designed, however, with that translational motivation.

“This was conceived very much in concert with the larger BrainGate* team, including neurosurgeons and neurologists giving us advice as to what were appropriate strategies for eventual clinical applications,” said Nurmikko, who is also affiliated with the Brown Institute for Brain Science.

Borton is now spearheading the development of a collaboration between EPFL and Brown to use a version of the device to study the role of the motor cortex in an animal model of Parkinson’s disease.

Meanwhile the Brown team is continuing work on advancing the device for even larger amounts of neural data transmission, reducing its size even further, and improving other aspects of the device’s safety and reliability so that it can someday be considered for clinical application in people with movement disabilities.

In addition to Nurmikko, Borton and Yin, the paper was also co-authored by Juan Aceros, an expert in mechanical engineering.

The National Institutes of Health/National Institute of Biomedical Imaging and Bioengineering and National Institute of Neurological Disorders and Stroke (Grant 1R01EB007401-01), with partial support from the National Science Foundation (Grants: 0937848) and the Defense Advanced Research Projects Agency (Contract: N66001-10-C-2010), funded the research.

*Caution: Investigational device. Limited by federal law to investigational use.

by David Orenstein

Brown researchers build robotic bat wing

The strong, flapping flight of bats offers great possibilities for the design of small aircraft, among other applications. By building a robotic bat wing, Brown researchers have uncovered flight secrets of real bats: the function of ligaments, the elasticity of skin, the structural support of musculature, skeletal flexibility, upstroke, downstroke.

PROVIDENCE, R.I. [Brown University] — Researchers at Brown University have developed a robotic bat wing that is providing valuable new information about dynamics of flapping flight in real bats.

The robot, which mimics the wing shape and motion of the lesser dog-faced fruit bat, is designed to flap while attached to a force transducer in a wind tunnel. As the lifelike wing flaps, the force transducer records the aerodynamic forces generated by the moving wing. By measuring the power output of the three servo motors that control the robot’s seven movable joints, researchers can evaluate the energy required to execute wing movements.

Wing of bat in life and lab
A robotic bat wing lets researchers measure forces, joint
movements, and flight parameters - and learn more about
how the real thing operates in nature.
Credit: Breuer and Swartz Labs/Brown University

Testing showed the robot can match the basic flight parameters of bats, producing enough thrust to overcome drag and enough lift to carry the weight of the model species.

A paper describing the robot and presenting results from preliminary experiments is published in the journal Bioinspiration and Biomimetics. The work was done in labs of Brown professors Kenneth Breuer and Sharon Swartz, who are the senior authors on the paper. Breuer, an engineer, and Swartz, a biologist, have studied bat flight and anatomy for years.

The faux flapper generates data that could never be collected directly from live animals, said Joseph Bahlman, a graduate student at Brown who led the project. Bats can’t fly when connected to instruments that record aerodynamic forces directly, so that isn’t an option — and bats don’t take requests.


Brown U. researchers build a "robatic" bat wing from Brown University on Vimeo.

“We can’t ask a bat to flap at a frequency of eight hertz then raise it to nine hertz so we can see what difference that makes,” Bahlman said. “They don’t really cooperate that way.”

But the model does exactly what the researchers want it to do. They can control each of its movement capabilities — kinematic parameters — individually. That way they can adjust one parameter while keeping the rest constant to isolate the effects.

“We can answer questions like, ‘Does increasing wing beat frequency improve lift and what’s the energetic cost of doing that?’” Bahlman said. “We can directly measure the relationship between these kinematic parameters, aerodynamic forces, and energetics.”

Detailed experimental results from the robot will be described in future research papers, but this first paper includes some preliminary results from a few case studies.

One experiment looked at the aerodynamic effects of wing folding. Bats and some birds fold their wings back during the upstroke. Previous research from Brown had found that folding helped the bats save energy, but how folding affected aerodynamic forces wasn’t clear. Testing with the robot wing shows that folding is all about lift.

Studying an animal with unique abilities
Over the years, Kenneth Breuer, an engineer, and
Sharon Swartz, a biologist, have developed a large
archive of bat data, from wind tunnels to field
studies and slow-motion video.

In a flapping animal, positive lift is generated by the downstroke, but some of that lift is undone by the subsequent upstroke, which generates negative lift. By running trials with and without wing folding, the robot showed that folding the wing on the upstroke dramatically decreases that negative lift, increasing net lift by 50 percent.

Data like that will not only give new insights into the mechanics of bat flight, it could aid the design of small flapping aircraft. The research was funded by the U.S. Air Force Office of Scientific Research and the National Science Foundation..

Inspired by the real thing

Bat wings are complex things. They span most of the length of a bat’s body, from shoulder to foot. They are supported and moved by two arm bones and five finger-like digits. Over those bones is a super-elastic skin that can stretch up to 400 percent without tearing. The eight-inch robot mimics that anatomy with plastic bones carefully fabricated on a 3-D printer to match proportions of a real bat. The skin is made of a silicone elastomer. The joints are actuated by servo motors that pull on tendon-like cables, which in turn pull on the joints.

The robot doesn’t quite match the complexity of a real bat’s wing, which has 25 joints and 34 degrees of freedom. An exact simulation isn’t feasible given today’s technology and wouldn’t be desirable anyway, Bahlman said. Part of why the model is useful is that it distills bat flapping down to five fundamental parameters: flapping frequency, flapping amplitude, the angle of the flap relative to the ground, the amount of time used for the downstroke, and the extent to which the wings can fold back.

Experimental data aside, Bahlman said there were many lessons learned just in building the robot and getting it to work properly. “We learned a lot about how bats work from trying to duplicate them and having things go wrong,” he said.

During testing, for example, the tongue and groove joint used for the robot’s elbow broke repeatedly. The forces on the wing would spread open the groove, and eventually break it open. Bahlman eventually wrapped steel cable around the joint to keep it intact, similar to the way ligaments hold joints together in real animals.

The fact that the elbow was a characteristic weak point in the robot might help to explain the musculature of elbows in real bats. Bats have a large set of muscles at the elbow that are not positioned to flex the joint. In humans, these muscles are used in the motion that helps us turn our palms up or down. Bats can’t make that motion, however, so the fact that these muscles are so large was something of a mystery. Bahlman’s experience with the robot suggests these muscles may be adapted to resist bending in a direction that would break the joint open.

The wing membrane provided more lessons. It often tore at the leading edge, prompting Bahlman to reinforce that spot with elastic threads. The fix ended up looking a lot like the tendon and muscle that reinforce leading edges in bats, underscoring how important those structures are.

Now that the model is operational, Bahlman has lots of plans for it.

“The next step is to start playing with the materials,” he said. “We’d like to try different wing materials, different amounts of flexibility on the bones, looking to see if there are beneficial tradeoffs in these material properties.”

- by Kevin Stacey

Brown Engineering Alumni H. David Hibbitt Ph.D. ’72 and Enrique Lavernia ’82 Elected to the National Academy of Engineering

Brown University engineering alumni H. David Hibbitt Ph.D. ’72 and Enrique Lavernia ’82 have been elected to the National Academy of Engineering (NAE). Hibbitt, founder and retired chairman of ABAQUS Inc. (now known as Dassault Systèmes Simulia Corp.), was honored for creation and development of the ABAQUS finite element code for nonlinear structural analysis and its worldwide dissemination. He is one of 11 new foreign associates elected.

“I have been truly fortunate in having so many talented colleagues who chose to join our efforts, so I view this award as coming to me as the representative of that team,” said Hibbitt. “It is a great honor for us all. It is the outcome of work by an amazingly strong team of applied mechanics people, mathematicians, and computer scientists, all working together to deliver the Abaqus software suite. Several others in that team also came from Brown Engineering, including Paul Sorensen ’71 Sc.M.’75 Ph.D.’77, Joop Nagtegaal Ph.D. ’73, David Berman ’84 Sc.M.’85, Mark Bohm ’84, and David Reynolds Sc.M.’91 Ph.D.’93.”

Lavernia, Dean of the College of Engineering, and Distinguished Professor of Chemical Engineering and Materials Science at University of California, Davis, was recognized for contributions to novel processing of metals and alloys, and for leadership in engineering education. He is one of 69 new members elected. The total U.S. membership is now 2,250 members and the number of foreign associates is now 211.

“This is a spectacular achievement for David and Enrique and we are extremely happy for them,” said Dean Larry Larson. “To have two alumni elected in one year from Brown is a wonderful accomplishment.”

Maurice Herlihy, professor of computer science at Brown, was also elected to the NAE this year for concurrent computing techniques for linearizability, non-blocking data structures, and transactional memory.

Michael Ortiz, who was a professor at Brown from 1984-1995 and is now a professor at California Institute of Technology, was elected for contributions to computational mechanics to advance the underpinnings of solid mechanics.

Election to the National Academy of Engineering is among the highest professional distinctions accorded to an engineer. Academy membership honors those who have made outstanding contributions to “engineering research, practice, or education, including, where appropriate, significant contributions to the engineering literature,” and to the “pioneering of new and developing fields of technology, making major advancements in traditional fields of engineering, or developing/implementing innovative approaches to engineering education.”

Hibbitt and Lavernia join an exclusive group of 12 Brown engineering alumni already in the NAE that includes: Walter J. Weber ’56 (elected 1985), William F. Allen ’41 (elected 1986), T. Dixon Dudderar PhD’66 (elected 1992), Wai-Fah Chen PhD’66 (elected 1995), George J. Dvorak (elected 1995), Marc S. Newkirk ’69 (elected 1997), Hratch Gregory Semerjian Sc.M.’68 Ph.D.’72 (elected 2000), Chain T. Liu Sc.M.’64 Ph.D.’67 (elected 2004), Robert M. McMeeking PhD’75 (elected 2005), Jean-Yves Parlange Ph.D.’62 (elected 2006), Alan I. Taub ’76 (elected 2006), and Ares J. Rosakis ScM’80 PhD’83 (elected 2011).

Ten current or former Brown engineering faculty members have been elected to the National Academy of Engineering, including Huajian Gao, Walter H. Annenberg Professor of Engineering, who was elected in 2012. Other members include: Vice President for Research and Otis Randall University Professor Clyde Briant (elected 2010), Subra Suresh (elected 2002), Professor Emeritus Alan Needleman (elected 2000), Professor Emeritus L.B. Freund (elected 1994), Rush C. Hawkins University Professor Rod Clifton (elected 1989), Joseph Kestin (elected 1982), James R. Rice (elected 1980), Daniel C. Drucker (elected 1967), and William Prager (elected 1965).

Robert Rome Named Associate Dean for Development and Planning at Brown University School of Engineering

Robert Rome has been named to the newly created position of associate dean for development and planning at the School of Engineering at Brown University. Rome began his duties on February 4 and is responsible for development of expanded master’s programs, growth of industry connections, planning for space growth of the School, communications, development, and diversity initiatives.

Rome comes to College Hill from the University of California San Diego, where he was the chief operations officer of the Department of Electrical and Computer Engineering. He brings a wealth of experience in development, student affairs, graduate program development, and financial management.

Rome holds a bachelor’s degree in psychology from American University and a master’s degree in education from the University of Pennsylvania.