Braingate array


Research published online first today, and in the May 17 2012 issue of Nature describes promising results of a clinical trial with tetraplegics (all four limbs paralyzed) that allowed the control of an external robotic arm (DEKA arm) using an embedded microarray in the brain, the BrainGate neural interface system.

One of the two study participants who had the array implanted 5 years ago, was able to use her mind to control a robotic arm and serve herself coffee from a bottle, 15 years after she became completely paralyzed & unable to speak.

The results from a team of researchers from the Department of Veterans Affairs, Brown University, MGH, Harvard Medical School (BrainGate Research Team) and the German Aerospace Center, Institute of Robotics and Mechatronics are promising.

The BrainGate system is the size of a small pill, and consists of a grid of 96 electrodes that are implanted in the motor cortex of the brain. By placing the grid next to the part of the brain that controls movement, neuronal activity associated with a movement can be translated into a computer command that drives an external robotic device.

The results reported by Leigh Hochberg MD, PhD & colleagues in Nature, offers hope to those paralyzed or who have limbs amputed, that in the future, innovations in neurotechnology may allow thoughts to control prosthetics or external robots.

Courtesy of Nature Video, you can watch how a paralyzed woman controls a robotic arm with her thoughts – this is an amazing video that is well worth watching:

How does it work, according to clinical trial subject T2:

I just imagined moving my own arm and the [DEKA] arm moved where I wanted it to go.

Subject S3 commented:

I think about moving my hand and wrist. I’m right handed so, it’s very comfortable and feels natural to imagine my right hand moving in the direction I want the robotic arm to move.

The challenge with this type of medical news is the danger of hype over hope, so to better put the results in perspective, I am delighted to have some expert commentary in the form of a guest blog post from D. Kacy Cullen, Ph.D, Assistant Professor at the University of Pennsylvania, Department of Neurosurgery Center for Brain Injury and Repair.

The Cullen laboratory at Penn applies Neural Engineering principles and technologies to the area of Neurotrauma, and is actively researching how to use neural tissue engineering-treatments to promote regeneration and restore function.

Commentary by D. Kacy Cullen, PhD 

This work is a natural extension/combination of the group’s previous work (1) involving human patients and computer cursor movements, and (2) non-human primates and robotic arm control.  So, this “next step” was anticipated, and in fact larger trials (involving various groups) are being initiated to investigate brain-based neural interface systems to drive the DEKA arm in subjects with limb loss (i.e. absent CNS damage/deficits).

The most compelling features to me were the decoding/training algorithm and the fact that one of the patients had her micro-electrode array implanted 5 years earlier.

Decoding/training: The use of signal filtering/thresholding in combination with open-loop (imagining and watching movement) and closed-loop (controlling the arm with visual feedback) recording/training was innovative and relatively efficient (31 min). However, in each case, the subjects had worked controlling arms previously (over years for S3 and 3 trials for T2).

A major challenge with recording ensemble neuronal activity in the motor cortex (or anywhere in the cortex) is signal attenuation and drift over time; so, each day/session typically requires re-training and re-calibration.

I would be curious to see how the subjects did in other independent trials – perhaps visual feedback can allow the user to “correct” the cortical inputs and hence reduce movement errors more rapidly in subsequent trials.  Nonetheless, it is remarkable that the subjects were able to manipulate the arm and drive it in a controlled and useful manner in a relatively short amount of time.

5 years: using an electrode array implanted 5 years earlier to control the robotic limb is very impressive.  The finding bodes well for the potential of this brain-based approach to yield useful cortical data chronically.

A major challenge is that over time the brain gradually rejects these non-organic electrodes, causing a build up of micro-scar tissue around the electrodes and a decreased neuronal density in the vicinity of the electrodes.

This process is partly due to mechanical mismatch between the electrode the brain causing inflammatory “micro-motion”.  This is likely exacerbated by subject motion/walking, which would not be an issue with the patients in this study but will be for the amputee study.

Nonetheless, this study noted “lower spike amplitudes and fewer contributing (active) channels” compared to earlier years, which is consistent with micro-scar tissue and fewer neurons close to electrodes.

Although this work is a natural next step, I do not want to trivialize the supreme competence, technical savvy, and attention to detail necessary to pull this off.  This group is highly competent and has the experience and skill to execute this very complex and multi-faceted neural engineering project to assist chronically disabled patients.


ResearchBlogging.orgHochberg, L., Bacher, D., Jarosiewicz, B., Masse, N., Simeral, J., Vogel, J., Haddadin, S., Liu, J., Cash, S., van der Smagt, P., & Donoghue, J. (2012). Reach and grasp by people with tetraplegia using a neurally controlled robotic arm Nature, 485 (7398), 372-375 DOI: 10.1038/nature11076