Brain-Machine Interfaces & Brain Plasticity

On Oct. 16, 2008, Julie Steenhuysen filed a report for Reuters entitled "Device helps monkeys move paralyzed wrists" describing a recent break through in fundamental research on brain-machine interfaces that considerably broadens avenues for the prosthetic control of limb movement. The findings are published in the journal Nature (Moritz and others, 2008). National Public Radio's Morning Edition provided an interview by Dan Charles entitled "Monkey Studies Could Help Paralyzed Humans" with the first author of the study. I have written about such interfaces in my post dated Jan 23, 2008.

The researchers at the University of Washington temporarily numbed nerves controlling arm movement in monkeys. Fine electrical leads were implanted into the area of cerebral cortex that controls limb movement known as motor cortex. The leads were used to record the electrical signals that nerve cells use to control skeletal muscle contraction. The signals were amplified, electronically transformed and fed into wire electrodes implanted into the muscles of the numbed arm. The monkeys learned to execute goal-directed movements with this limb. The results constitute a mile stone proving both the applicability of the electronic interface and the versatility of the motor system to utilize the new extraordinary tool in a meaningful fashion.

The nerve cells in our central nervous system that innervate the skeletal musculature are known as motor neurons. When peripheral nerve injury severs their axons, that is the nerve fibers that establish the connections with the muscle fibers, motor neurons can regenerate the disrupted connections. During this period, the cells are subjected to remarkable alterations. A glial reaction ensues in their vicinity. In my own experience, strong signs of the glial response can be detected on histological tissue sections within four days after nerve injury. The signs are visible on this micrograph from a transverse section through the brain stem of a rat.

microglia, courtesy of J.A. McKanna

The hypoglossal nerve, that is the twelfth cranial nerve innervating the muscles of the tongue, was damaged on the right side (left in the micrograph). The bodies of nerve cells are stained blue in the micrograph. Microglia are stained black. The cell bodies of the axotomized motor neurons (asterisk) are located left of the center of the section in an area called hypoglossal nucleus. Microglia (arrowhead) are gathered in great number among the axotomized motor neurons and wrap themselves around their bodies (arrow), detaching incoming nerve contacts known as synapses that convey command and control for muscle contraction from the fore brain. The motor neurons undergo chromatolysis and increase protein synthesis. David Bodian described the cellular changes using electron microscopy in great detail in the Johns Hopkins Hospital Bulletin (Bodian, 1964). Blinzinger and Kreutzberg (1968) were the first to identify the cells that insert themselves between the motor neuron and the synapses as microglia. After roughly two months the glial reaction ceases, the synapses re-attach to the cell bodies, and the motor neurons regain much of their original appearance. Major histo-compatibility complexes have been identified as one major group of signal molecules that control the observed glial and motor neuron responses to axotomy (Oliveira and others, 2004).

The ability of motor neurons to re-establish disrupted muscle innervation is a fascinating example of our brain's ability to recover from injury. However, it is important to note that the repair is imperfect. The novel innervation commonly remains below original strength and the endings of the motor neurons may not succeed in finding their original muscle fibers (Madaschi and others, 2003). Intriguingly, the nerve cells in the central nervous system are able to adjust to the altered peripheral innervation. Sprouting of novel connections has been proposed as mechanism (Fujito and Aoki, 2002). In fact, the plasticity of the motor system is so great that animals reportedly learn meaningful limb movements even after the surgical cross of nerves controlling antagonistic muscles [Sperry, 1941 (reviewed by Todman, 2008)].

Taking this enormous flexibility of the motor system into consideration, the directed arm movements of the interfaced monkeys Moritz and others (2008) observed may not entirely come as a surprise. Doubtlessly, the technology to transform the nerve cell signals recorded in the cerebral motor cortex into meaningful stimuli for the arm muscles is a daunting achievement. However, it is important to emphasize that the success of this method ultimately relies upon the nerve cells that alter their electrical discharges in order to produce the desired movement. As pointed out on the National Public Radio broadcast, the fascinating discovery is the rapidity with which the nerve cells learn to direct a movement under extraordinary experimental conditions. The question remains to be answered whether special cells or a special ensemble of cells is needed to produce fine-grain limb control.

Addendum

• On Feb. 10, 2009, Pam Belluck reported in her post entitled "In New Procedure, Artificial Arm Listens to Brain" for The New York Times on a promising variation of this idea published in The Journal of the American Medical Association (JAMA 301(6):619-628). With the new procedure, Kuiken and others (2009) planted wire electrodes over functional muscle groups that a patient with a lost limb can control. The electrical nerve signals recorded from the electrodes when the patient is using the underlying muscles are subsequently employed to steer a prosthesis replacing the missing limb. With practice, the patients learn to substitute the contractions of the intact muscles with prosthetic limb movements to the extent that they feel the limb manipulated when the skin over the muscles is touched (02/11/2009).

References

• Blinzinger K, Kreutzberg G (1968) Displacement of synaptic terminals from regenerating motoneurons by microglial cells. Cell Tissue Res 85: 145-157.
• Bodian D (1964) An electron-microscopic study of the monkey spinal cord. I. Fine structure of normal motor column. II. Effects of retrograde chromatolysis. III. Cytological effects of mild and virulent poliovirus infection. Bull Johns Hopkins Hosp 114: 13-119.
• Fujito Y, Aoki M (2002) Distal forelimb cross-innervation effectively induces formation of corticorubral synapses. Neuroreport 13: 2121-2124.
• Kuiken TA, Li G, Lock BA, Lipschutz RD, Miller LA, Stubblefield KA, Englehart KB (2009) Targeted muscle reinnervation for real-time myoelectric control of multifunction artificial arms. JAMA 301: 619-628.
• Madaschi L, Di Giulio AM, Gorio A (2003) Muscle Reinnervation and IGF-I Synthesis Are Affected by Exposure to Heparin: An Effect Partially Antagonized by Anti-Growth Hormone-Releasing Hormone. Neurochem Res 28: 163-168.
• Moritz CT, Perlmutter SI, Fetz, EE (2008) Direct control of paralysed muscles by cortical neurons. Nature 456: 639-642.
• Oliveira AL, Thams S, Lidman O, Piehl F, Hökfelt T, Kärre K, Lindå H, Cullheim S (2008) A role for MHC class I molecules in synaptic plasticity and regeneration of neurons after axotomy. Proc Natl Acad Sci U S A 101: 17843-17848.
• Sperry RW (1941) The effect of crossing antagonistic muscles in the hind limb of the rat. J Comp Neurol, 75: 1-19.
• Todman, D (2008) History of Neuroscience: Roger Sperry (1913-1994), IBRO History of Neuroscience

Related Posts

• Wo(Man) & Machine

(Wo)Man & Machine

I once was privileged to witness a controlled wrecking on a college campus. The goal of the project was not to destroy the whole structure, a three-story laboratory building, but only the two-storied annex. This "surgical cut" was executed using a hulky excavator armed with a oversized pneumatic jackhammer. The hammer was shaped like a giant tooth with which the operator probed the structures' beams. The operator gingerly moved the tooth about the building, gently probing here and there, constantly searching for the critical junctions that, once the jackhammer was unleashed, would only let the annex collapse. The operator approached the job with extreme diligence. The pneumatic hammer was activated with utmost nimbleness, leaving the observer with the impression that the person at the controls was able to sense the softening of the structure through the machine's tooth. Although the procedure was carried out with urgency, progress was excruciatingly slow. No mistakes were allowed. Blows were dealt with a sensitive touch. Despite the formidable challenge, the annex eventually turned into a pile of rubble while the main building remained unharmed. In the end, a lady with a big happy grin stepped from the cab to the ovations of a small crowd of academic onlookers that had gathered.

The operator's performance impressed so profoundly, because she appeared to control the excavator as if it were part of her body. In support of this idea, the machine mediates impact and vibration, providing meaningful tactile and proprioceptive feedback to the operator. Combined with visual and auditory cues, this feedback may produce an accurate perception of the machine's precise dimensions and forces of engagement can be gauged with the necessary accuracy. Reinforced by the continued use of successful strategies, connections between nerve cells strengthen, forming new networks that ultimately represent the machine in the mind as an extended part of the body.

Such plasticity of nerve cell networks is fundamental to the demonstration reported by Sandra Blakeslee in her article entitled "Monkey’s Thoughts Propel Robot, a Step That May Help Humans" in The New York Times on Jan. 15, 2008, that a monkey in the U.S. could steer a robot's walk in Japan with neuronal action potentials, that is electric nerve cell signals, recorded from her brain. The work was carried out in Miguel Nicolelis' laboratory at Duke University. The monkey was trained to walk on a treadmill while watching on a video screen the back of the robot walking on a similar mill, seemingly ahead of her. Microwire electrodes were implanted through an opening in the skull into the monkey's somatic sensory and motor cortex to record nerve cell signals. Signals that control leg movements were transmitted to Japan via the internet and fed to the robot's computer controlling its walk. The monkey was rewarded with treats for keeping the robot walking on the belt. Eventually, she managed to keep the robot on the move, while giving herself a rest.

The computer monitor shown in the Reuters video at 0:18 minutes:seconds displays electrical nerve cell activity in the monkey's cortex acquired with a Plexon's multichannel acquisition processor. The large window on the right side depicts recordings from 128 electrodes arrayed in a matrix of 16 x 8 channels. Each electrode picks up electrical spikes from a number of cells. However, particularities in spike shape can be used to identify individual cells. The sorting of nerve cells by wave form is displayed in the windows on the left. The upper window depicts the spikes recorded from one electrode. The red, green, yellow and blue traces identify the spikes of four nerve cells. The isolated spike wave forms are shown separately in the window below. The upper window is shown enlarged at 0:23. The electrode channel from which the recordings were taken is framed in red on the left border of the enlarged matrix window at 0:28. Only the nerve cell signals essential to the control of the monkey's legs were used to control the robot's walk.

The success of this demonstration doubtlessly constitutes a formidable achievement of science and engineering. Yet, the greatest accomplishment resides in the monkey's brain. Similar to the big machine operator, the monkey experienced positive reinforcement when the robot walked successfully. During training, nerve cells in the monkey's motor cortex must have modified their connections such that their signals could be correctly interpreted by the robot's computer to keep the machine on the treadmill. As Eugen Herrigel so befittingly described in Zen in the Art of Archery the arrow, i.e. the nerve cell signals, and the bull's-eye, i.e. the robot, must fuse into one to get the job done.

The technology of controlling computers with nerve cell signals has been already applied to humans. Kennedy and Bakay (1998) demonstrated that a paralyzed patient could intentionally move a cursor on a computer monitor with nerve cell signals recorded from an electrode implanted into motor cortex. Eliminating the need for an opening in the skull, Birbaumer and others (2006) showed that recordings of small electrical currents on the scalp can be used to work a word processor. Nerve cells may even be able to control computers directly one day. Peter Fromherz and colleagues successfully grew endings of nerve cells onto silicon wafers in tissue culture and provided evidence that the nerve cell signals influenced the flow of electrons in the semiconductor (Fromherz and others, 1991). The implementation of this idea in vivo is not entirely utopian. Damaged peripheral nerve cell fibers are known to regenerate and may establish functional connections (Melzer and Smith, 1995).

Addenda

• Richard Allen Greene reports in his post entitled "Brain-Twitter project offers hope to paralyzed patients" on CNN today on exciting progress that Adam Wilson and Justin Williams at the University of Wisconsin have made in developing a method with which electrical signals of nerve cells in the brain recorded from the scalp can be used to compose a message on Twitter, opening up a new avenue for paralyzed people to communicate (04/23/09).
• On National Public Radio's All Things Considered today, Michele Norris anchored a segment entitled "Your Brain On Twitter: No Hands Necessary" on brain wave-controlled Twitter messages (04/24/09).
• This short, but impressive video clip below superbly corroborates my observation. Bionic legs will assist us on our first steps into a very useful direction (07/18/10).

• Today, Diane Rehm interviewed Miguel Nicolelis on her show with the title "Miguel Nicolelis: 'Beyond Boundaries'". Dr. Nicolelis is a lead investigator in this research and discusses his fascinating insights (03/16/11).
• Nicolelis and others (2011) published the proof of concept for nerve cell activity-controlled movements of a virtual limb combined with feedback through electrical microstimulation of nerve cells in somatic sensory cortex in this week's issue of Nature (10/06/11).

References

• Birbaumer N, Weber C, Neuper C, Buch E, Haapen K, Cohen L (2006) Physiological regulation of thinking: brain-computer interface (BCI) research. Prog Brain Res 159: 369-391.
• Fromherz P, Offenhäusser A, Vetter T, Weis J (1991) A neuron-silicon junction: a Retzius cell of the leech on an insulated-gate field-effect transistor. Science 252: 1290-1293.
• Kennedy PR, Bakay RA (1998) Restoration of neural output from a paralyzed patient by a direct brain connection. Neuroreport 9: 1707-1711.
• Melzer P, Smith CB (1995) Whisker follicle removal affects somatotopy and innervation of other follicles in adult mice. Cereb Cortex 5:301-306.
• O’Doherty JE, Lebedev MA, Ifft PJ, Zhuang KZ, Shokur S, Bleuler H, Nicolelis, MAL (2011)
  Active tactile exploration using a brain–machine–brain interface. Nature doi:10.1038/nature10489.