Human Brain to Machine Interface May Now Be Feasible

Laurie Barclay, MD

May 06, 2004

May 6, 2004 — Directly using human brain neuronal activity to operate external neuroprostheses may now be feasible, according to a presentation on May 4 at the 72nd annual meeting of the American Association of Neurological Surgeons held in Orlando, Florida. This work could potentially benefit patients with quadriplegia or other focal neurological injury who are unable to use their extremities because of a breakdown in connectivity between their limbs and brain motor centers.

"For all kinds of motor training, such as riding a bicycle, people incorporate an external device into their schema, and the process becomes subconscious," senior author Dennis Turner, MD, MA, a neurosurgeon at Duke University in Durham, North Carolina, told Medscape. "We will build on that phenomenon in our human studies."

Earlier studies revealed that monkeys could use brain signals from hundreds of neurons chronically implanted with high-density microelectrode arrays to control a neuroprosthetic robotic arm.

In this study, 11 volunteer patients with Parkinson's disease had ensemble recordings of neuronal activity from a novel 32-channel electrode array implanted in the subthalamic nucleus and thalamic motor regions during surgical placement of deep brain stimulators. There were no intraoperative complications.

At the same time, the patients played a hand-controlled video game for up to 10 minutes, in which they were placed in a semi-sitting position before a video screen and instructed to adjust their hand-gripping force on a squeeze ball to reach a target level of force as quickly as possible. Shoulder and arm contractions were prevented by positioning the limb on an armboard so that the only motion tested was the patient's gripping force generated by the hand.

During 23 recording sessions, the investigators simultaneously recorded up to 55 neurons. After optimal training, synchronously recorded neuronal activity predicted motor activity (r = 0.64; r2 = 0.68). Larger neuronal ensembles produced more accurate motor predictions, with 61% of neurons from the subthalamic nucleus and 81% of neurons from thalamic motor areas varying with gripping force.

These recorded signals contained sufficient information to reliably predict hand motions, and therefore to accurately control an external robotic prosthesis. This was especially encouraging because there were only about five minutes of data per patient, which included about one or two minutes to train the patient to perform the task. As clinical testing progresses and electrode arrays are implanted for a long period of time, it would be easier to achieve a workable control system for external devices, according to the authors.

Unlike the monkey studies, in which recording was from the cortical surface, the human studies used recording from subcortical structures. Although both locations may provide viable options for sampling neuronal information to control a prosthetic device, subcortical electrodes offer certain advantages. The subcortical areas are denser, with more cells to record from in a smaller area; they filter all the signals for motor control before they reach the final cortical output; and they are deeper, so electrodes implanted there are more stable than cortical electrodes.

In addition to a neuroprosthetic hand, potential applications of this technique include a neurally controlled wheelchair or keyboard, or even a speech synthesizer for patients with stroke or amyotrophic lateral sclerosis.

"Patients who don't have use of their arm still show in MRI studies that the control centers in the brain are working normally [and become active] when they are asked to imagine moving their arm," Dr. Turner says. "We have good hope that the neurons in those centers can still provide the same signals, even though the arm isn't physically working."

Although the investigators have applied for approval to begin implanting experimental electrode arrays for long-term use in quadriplegic patients, they caution that many years of development and clinical testing must ensue before any neuroprosthetic devices are clinically available.

This work is in press in the July 2004 issue of Neurosurgery. The Defense Advanced Research Projects Agency and the National Institutes of Health supported this study.

AANS 72nd Annual Meeting: Abstract 825. Presented May 4, 2004.

Reviewed by Gary D. Vogin, MD