Brain-Machine Interface Pioneers Participate in Workshop on Clinical BMIs 

Staff report 

Neural engineers, clinicians, and neuroscience researchers came to Houston, TX last month for the 2013 International Workshop on Clinical Brain-Neural Machine Interfaces. The first-ever event was sponsored by the University of Houston, the Methodist Hospital Research Institute, and The Institute for Rehabilitation and Research, with major funding from the National Institutes of Health and the National Science Foundation. 

Conference chair Pepe Contreras-Vidal from the University of Houston set the stage for the three-day event with a challenge to participants to create a roadmap for moving from proof of concept to safe, reliable use in patients. He called attention to a previously produced European roadmap on BMI development as a starting point. 

Kip Ludwig from the NIH advised attendees to consider all the stages that would be required to produce a commercially viable product, from early feasibility trials to regulatory approval and reimbursement. He recommended establishing metrics for determining the risk/benefit equation and stressed that there would likely be several iterations in the process. 

Issues of terminology and definitions emerged early during the event. John Donoghue from Brown University challenged conventional notions of BMI “invasiveness,” reminding attendees that in many cases, ECoG electrodes mounted on the surface of the cortex could be more damaging than penetrating microelectrodes normally considered to be at the highest level of invasiveness.  

Jack Judy from DARPA urged attendees to not limit their thinking to “brain” machine interfaces but to consider interfaces to peripheral nerves and the spinal cord. Judy is program manager of the Reliable Neural Interface Technology program at DARPA’s Microsystems Technology Office. He reported that in contrast with users of lower-extremity prostheses, upper-extremity amputee servicemembers have not been as happy with their prosthetic devices. 

In part because of the complexity of the devices and the learning curve required, it is taking a long time for users to achieve mastery of their systems, although those that do tend to be satisfied with them. Judy said that it is critical that the systems get a reliable signal from the user and decode that signal faithfully. Judy described three different approaches with peripheral neural interfaces, including targeted muscle reinnervation techniques perfected at the Rehabilitation Institute of Chicago, the FINE electrode developed at Case Western Reserve University, and the Implantable MyoElectric Sensor developed at the Alfred Mann Foundation.  

Keynote speaker Eberhard Fetz from the University of Washington gave an update on bidirectional interactions between the brain and implantable devices. He argued for recurrent connections in BMI devices and gave examples from some of his monkey work of how the brain adapts to consistent sensorimotor conditions and learns to incorporate a recurrent device using synaptic plasticity. 

Fulfilling the international mission of this workshop, presenters from EPFL in Switzerland and several Japanese institutions shared some of their results using BMIs with patients, including several who were locked-in. Jose del R Millan from EPFL encouraged attendees to adopt a user-centered approach. He cited the cochlear implant as an example of a device that could last for a long time within a user’s body. 

Leigh Hochberg from Mass General and Brown offered his top 10 characteristics for BMIs. A device should be safe, work 24/7, not require a caregiver, offer real-time control over limbs or assistive devices, require minimal concentration on the user’s part, restore communication at the speed of speech or typing, be acceptable aesthetically, last several years between battery changes, be reimbursed by insurance, and, of course be commercially available. Hochberg described work with eight BrainGate participants, including two ongoing users. The device has now accumulated over 5000 patient-days of service. 

Hunter Peckham from Case Western Reserve University shared progress on his work at the Cleveland FES Center. His team has developed a fully implantable, modular, and scalable neuroprosthesis system, including a bilateral upper-extremity device with myoelectric control for a C6 quadriplegic. Peckham said the system can control individual digits and multiple grasp patterns and suggested that flexibility might be more than can be derived from cortical control systems. 

Ron Diftler from NASA gave attendees an overview of the Robonaut 2 system, a highly dextrous anthropomorphic robot being developed in conjunction with General Motors. NASA is also developing an exoskeleton system that offers 42 degrees of freedom plus hand grasp capability. 

In addition to the technology sessions, there was a worthwhile session devoted to users of neuroprosthetic systems, who offered their feedback on some of the exoskeleton systems they had tried. While in general, the users were not entirely satisfied with the systems, each of them appreciated the availability of the devices and expressed their desire for improved systems. 

The users wanted to see devices that were quieter, more responsive to their inputs in real-time, and lighter weight. Most of the users brushed aside issues of risk—“The only risk for us is disappointment”—said one. Another user, however, said he would have to think long and hard before agreeing to a surgically implanted FES system. 

Exhibitors at the workshop included Tucker-Davis Technologies, Plexon, Blackrock Microsystems, Ripple, NeuroConn GmbH and Brain Products GmbH from Germany, G.tec Medical Engineering GmbH from Austria, and Neuroelectrics from Spain. Also exhibiting were exoskeleton manufacturers Rex Bionics from New Zealand and Parker Hannifin’s Human Motion & Control unit, which makes the Indego system. 

Other sponsoring institutions included Rice University, Tecnologico de Monterrey from Mexico, and Universitat Tübigen from Germany. 