Until recently, mechanical performance has been the limiting factor in how well a device can restore function to a patient, both for artificial limbs and most certainly for gait assistive orthoses. Advances over the last decade in the sophistication of mechanical operation of these devices has given patients such a lift in quality of life that they appear to approach near normal function – when viewed from the outside.

Prosthetic devices have benefitted most from this development boom but gait assistive orthoses (or exoskeletons as they are generally referred to now) are rapidly catching up, with some systems now allowing paraplegic patients to walk fully independent of human assistance – at least, in the rehabilitation training centre.

From the perspective of the patient however, these devices can still act and importantly feel very ‘other’ from themselves, as the interaction between the patient, the control system and the device mechanics is still very far from natural due to an all-too-common mismatch between control input and device action.

Using these devices also remains significantly taxing on the patient in terms of both physical effort which results in muscle fatigue from continuous exaggerated actions, and cognitive load from the need to remember and recall a growing number of unnatural control actions. For exoskeletons the effort is still very visible to see, but in modern prostheses this effort is now hidden behind slick pre-programmed mechanical actions.

While physical exertion can be addressed through mechanical power within the device limited only by on-board power capacity, physical and cognitive load can only be alleviated by careful control system design – and fidelity of control inputs.

Accurate neuro-muscular signal acquisition and interpretation is obviously key to enabling a high-fidelity control system. However, limitations of current sensors used in these devices places a ceiling on signal fidelity, so impacting the level of fine control able to be achieved. Beyond this ceiling is where the patient must consciously make up the difference in ability – the physical and cognitive load.

Limits for prosthetics and orthotics

For prosthetics, surface electromyography (EMG) has become the established sensing system for modern powered prostheses – because myoelectric sensors are cheap, well characterised, and can be incorporated within most socket types. However, there is a fundamental limit to the level of fidelity that can be obtained from surface EMG, due to attenuation of neuro-muscular signals, as well as dispersion of individual signal sources, through the skin.

And even with advanced techniques like targeted muscle re-innervation, the actions required to generate the EMG signals needed to drive a myoelectric prosthesis are still convoluted and unnatural, adding physical and cognitive load. This is further exaggerated with the use of ‘high functioning’ pre-programmed multi-grip devices, that require specific contraction pattern inputs to select the grip form, before being able to control the device in its simpler open/close mode.

For orthotics patients with neuro-muscular impairment or paralysis, surface EMG is simply not an option. Current commercial systems must rely on manual use of control keys to drive the hardware, with the exoskeleton running pre-programmed movement patterns to restore ambulation. This removes completely any semblance of natural or intuitive control from the device.

So, while reducing the physical exertion of the patient, active exoskeletons can actually increase the physical and cognitive load, by resorting to a control system that completely relies on conscious action – while at the same time trying to maintain balance and position within the device, and also likely using supporting crutches as well.

The fundamental consideration to the development of next generation prosthetic and orthotic devices is the simple premise – if surface EMG is not able to provide the required fidelity of control inputs, then where can ‘better’ control signals be obtained from?

This can only be the nervous system. This is where the skin, being seen increasingly as the barrier inviolate, is now the principal barrier for the future development of prosthetic and orthotic devices.

A barrier of more than one level

The skin has more than just the perception of a physical barrier to overcome – there is a psychological barrier in both the patient and the physician to moving to an implanted system. Implants have always had this mental barrier, and that barrier is spelt ‘S U R G E R Y’.

This psychological aspect is a little strange when it comes to prosthetics – the therapeutic pathway already includes surgery to perform the amputation, so access to the nervous system is readily presented and salvage of remaining nerves is part of standard procedure.

It is a little more understandable for neuro-muscular impairment patients. After what reconstruction surgery is required to repair the damage that caused the impairment, further surgery is considered best avoided, in order to give the remaining nerves the chance to restore and reroute to what pathways may still be viable.

Either way, in prosthetics and orthotics, surgery has always been considered as something that is a last resort and to be avoided if possible. The general drive in the medical field to minimal / non- invasive techniques and non-surgical therapies is laudable of course and any development of a therapy that reduces patient trauma in the course of that therapy is one that should be pursued – but not blindly.

This is because, in the field of prosthetics and orthotics, this direction is actually detrimental to the ability to restore higher functional levels of these devices. The only place that higher fidelity control signals can be obtained for these patients is from within the body.

Neural integration to control prosthetic and orthotic devices is a great leap forward, in capability, control signal fidelity – and, yes, implementation complexity. Tapping into the nervous system is currently, no question, difficult – but only because it has not had the decades of development and investment that surface sensing systems have received.

We are by no means starting from square one, however.

Beyond surface EMG

Targeted muscle re-innervation is a technique that takes a first step across the barrier, by recognising that good control fidelity must come from the original nervous connections to specific actions. It has established the surgical techniques required to identify and delineate the fascicles in peripheral nerves related to specific muscle actions, and their re-attachment in redundant muscle segments allows surface EMG to be used for patients with levels of amputation where it would otherwise be precluded.

But if you’re going through the surgical effort to reconnect these nerves, why not connect directly to a sensing electrode and by-pass the detriments of trying to sense through the skin?

Osseointegration fixation pins take another step across the barrier, offering a percutaneous conduit for neuro-muscular signals. This direct connection through the skin ensures both signal integrity and source discrimination as it crosses the barrier, so maintaining signal fidelity.

It allows a much higher channel count than would be available from surface sensing, simply due to the far smaller interface footprint required for implanted neural electrodes.

It also offers a mounting option that alleviates a considerable number of issues found with prosthetic sockets, such as day-to-day fit variability, range of motion limitations as well as basic issues of long-term comfort and device weight – a socket can contribute up to half the weight of a prosthesis.

Advances in material technology have allowed these fixations to become fully integrated into the body, ensuring long term strong and stable connections.

These fixations also raise the intriguing possibility of having also orthotic devices directly integrated in the patient’s skeleton, bypassing the skin and so the associated similar issues of the skin-device interface that prosthetic sockets suffer, as well as the soft tissue energy losses experienced as the orthosis drives the patient. However, this application is yet to been seen outside of rigid orthopaedic reconstruction support cages.

The above techniques have the base assumption that the neuro-muscular pathways required to control high function prostheses and orthoses are still available to tap into. Where the neuro-muscular system is absent or damaged, only a direct neural connection is able to restore the signal pathway. The issue then becomes how far up the neural pathway does the connection need to be made?

But this could be at any point in the nervous system, from distal peripheral nerves all the way up into the brain, depending on the level of amputation or nerve damage.

Nerve interfaces

Peripheral nerve electrode interfaces are a long-established technology and have little further complication than any other sub-dermal procedure. In their application, paddle, cuff and micro-wire electrodes have well proven the capability to sense signals from the nervous system over extended periods of time. The only limitation on these sensors is the size they can be manufactured and deployed at.

Micro surgical techniques for nerve reconstruction allow the manipulation required to perform the connection, but electrode embodiments of the appropriate scale required to target specific peripheral nerve fascicles, on the order of sub-millimetre down to a couple of hundred microns, are not typically available outside of research programs.

At higher levels in the peripheral nervous system and certainly in the spinal cord, it becomes more an issue of being able to dissect through the compacted nerve bundles to locate and identify the appropriate fascicles of the specific neuro-muscular pathway. This is where the mental and technical barrier of surgery escalates and dominates, as this is delicate and time-consuming work with severe implications of any collateral damage.

So, what about at the level of the brain? If the prospect of surgery at any level has a high psychological barrier, adding ‘Brain’ in front of it only magnifies this mental hurdle. But this is where the intent of neural signals is at its purest and the location of the driving centres for the various neuro-muscular pathways best defined – the motor cortex is one of the most well mapped areas of the brain.

For high-level quadriplegics, where the nerve pathway damage has left all peripheral or spinal connection points unusable, this is the only place remaining where control signals can be sensed from.

Developments in flexible electronics and micro-fabrication techniques now allow unparalleled access to the signals previously hidden away in the brain, with sensor point densities finally reaching the required scale to extract distinct fine motor action signals from the cortical surface.

Neural integration

Without this deeper, more direct connection to our patients, the devices prescribed to them will always remain as something other to themselves, with a fundamental disconnection irrespective of the level of mechanical sophistication able to be performed.

The potential of neural-integrated implants in prosthetics and orthotics is no less than the restoration of natural and intuitive control of these devices. This will cement the fundamental step-change in the integration of these devices into the perception of ‘self’, that the contrived, exaggerated, push-button control inputs required by current systems simply cannot address.

It also holds the promise of returning sensory information back into the nervous system, finally allowing the device to become part of the person, rather than just an external apparatus.

Neural-integrated control systems must be the next stage of prosthetic and orthotic device development, otherwise they shall remain only ever as assistive devices, rather than truly restorative.

We need to get under the skin of our patients, in order to give them back their bodies.

Expert authors