Peripheral Nerve Interfaces, Brain Computer Interfaces (BCIs), and Muscle Stimulators
oxford neural interfacing
The Oxford Neural Interfacing Group led by James FitzGerald is a collaboration between Neurosurgeons and Biomedical Engineers. The group is dedicated to designing, developing and testing new ways to connect the human nervous system to electronic devices in order to restore function or aid rehabilitation after injury. We work on peripheral nerve interfaces, brain computer interfaces (BCIs), and devices to stimulate weak or paralysed muscles.
Peripheral nerve interfacing
Very sophisticated robotic prostheses have been developed to help patients unfortunate enough to have lost an arm or hand. They are capable of a wide range of movements closely mimicking those of a real limb, but their full potential cannot be realised because the means by which the user has to control them is much more primitive.
Prosthetic limbs currently employ "myo-electric" control. Electrodes on the skin surface detect activity in muscles in the amputation stump, which the user learns to twitch voluntarily to instruct the prosthesis to, for example, open or close the hand. This method allows control of only one or two movements at any given time, users are unable to produce graded force, and the control always requires conscious effort.
To unlock the capabilities of advanced robotic prosthetics, and allow the user to fully and effortlessly control them by thought alone, direct connection to the severed nerves in the amputation stump is needed. We are using a novel type of nerve interface based upon growing nerve fibres from the nerve stump into a bundle of closely packed narrow channels (typically a tenth of a millimetre in diameter) that contain recording electrodes. The design allows both signal amplification and noise reduction, leading to reliable performance in recording nerve signals.
Some key hurdles have recently been overcome. The body reacts to medical implants by coating them with a thin layer of scar tissue, and for an interface this is a major problem as the scar layer gets in between nerve fibres and the recording electrodes. We have developed a way to suppress this scarring, using the slow release of miniscule quantities of anti-inflammatory medication from the device itself, a technique called drug elution. This will ensure stable device function for the long term. Our present focus is on methods of decoding signals recorded from the interface to turn them into servo control instructions for a prosthetic limb.
Brain computer interfaces
Approximately 500 people per year in the UK suffer a spinal cord injury in the upper neck that leaves them paralysed in all four limbs. This leads to a life of near total reliance on others, and any technological advance that can provide even modest degrees of independence can have a substantial impact on long term quality of life.
Limb movements are controlled by a specific region of the brain called the motor cortex. Brain cells here send long nerve fibres down through the spinal cord, which carry the nerve signals from the brain that control movement. These fibres are cut by the injury, but nerve cells in the brain survive and remain active. Brain-computer interfaces are devices designed to record electrical activity from the parts of the brain that used to control the paralysed areas, in order that the recorded signals can be used to control devices like electric wheelchairs or stimulate paralysed muscles.
Most current approaches to doing this use signals recorded from either electrodes on the scalp surface (electroencephalography or EEG), or electrodes implanted into the brain itself. The former are noninvasive which makes them safe but insensitive, while the latter can record very detailed information but have limited longevity of function and eventually damage the brain.
We are using an intermediate solution where electrodes are built into a flat sheet that is positioned on the surface of the brain, but do not penetrate it, a technique called electrocorticography (ECoG). ECoG affords far higher spatial resolution and sensitivity than external EEG based approaches, while avoiding electrode insertion into the brain itself.