“Patients who had lived with a spinal cord injury for twenty years, and had never felt anything in their arms, could still feel electrical stimulation deep in their muscles — proof that something was still getting through the injured spinal cord”.
What if simply thinking about moving your arm could help retrain a damaged spinal cord — or bypass it altogether? That’s the question at the heart of Aleksandra Vuckovic’s work. Professor of Neurotechnology and Rehabilitation Engineering at the University of Glasgow and Co-Director of Research at the Scottish Centre for Innovation in Spinal Cord Injuries, Vuckovic develops closed-loop brain-computer interfaces that turn brain activity into movement and pain relief for people living with spinal cord injury. We spoke with her on the sidelines of the IFESS conference, where she was one of the invited speakers.
Your research centers on closed-loop sensory-motor systems driven by brain-computer interfaces (BCIs). Could you elaborate on that?
In simple terms, we record brain activity directly and help people use it — rather than speech or muscle activity — to control devices. It can be a little difficult at first, but I think it’s ultimately the most intuitive approach, because you skip all the intermediary steps in between. It becomes almost natural: when you move your arm or leg in everyday life, you don’t verbalize the action or rely on any intermediary, you simply think of it, and it happens.
There are two reasons behind this approach. The first is that we are training the brain and the remaining pathways from the brain, through the spinal cord, to the muscles: activating the brain while imagining movement ultimately activates the downstream pathways towards the muscles as well. The second applies when the injury is complete and no muscle activation is possible at all — in that case, we bypass the injured section of the spinal cord entirely and connect the brain directly to an actuator, whether that is a muscle driven by an electrical stimulator, or an orthosis.
So, the strategy depends on the population we’re working with. In people with incomplete injuries, where the goal is recovery, we try to activate whatever residual capacity remains in the injured spinal cord. In people with complete injuries — which I work with less, since we don’t see that many truly complete cases — the goal shifts to control: using the BCI to operate a stimulator or an orthosis by bypassing the injured pathway altogether. So, for rehabilitation and recovery, we aim to activate as much of the injured spinal cord as possible; for control in chronic, complete injury, we aim to bypass it.
Current statistics show that most spinal cord lesions today are incomplete — increasingly incomplete rather than complete. Is that right?
That’s very true. And even people who are clinically classified as complete often still retain some residual connectivity — which genuinely surprised me when I first started this work. When we applied electrical stimulation to the muscles and asked people to imagine movement, we would visit patients at home, often out in the Highlands, since we’re based in Scotland. Many of them had been inactive for twenty years, had never tried a brain-computer interface, and had never felt anything in their arms before. They told us they could feel the electrical stimulation deep in their muscles — though it took a stronger stimulus than usual to get there — and I could see corresponding brain activity appear in response to that stimulation. So clearly, something was still getting through the injured spinal cord. That’s consistent with what’s reported in the literature, and it’s exactly what I observed in my own studies as well.
Aleksandra, you work on both restoring motor function and alleviating chronic pain in people with spinal cord injury. From a neurophysiological standpoint, do these two goals share a common underlying mechanism, or do they require different approaches?
Yes and no — they do share a common space. Disturbed sensation, over the long term, is what tends to give rise to chronic neuropathic pain. But we now know that neuropathic pain involves cortical areas beyond the “pain matrix” — it also involves the motor cortex, the region that controls movement. That’s where the link between the two lies.
When we treat pain, we’re essentially trying to relax the motor cortex; when someone attempts movement, they’re activating it. So, in a sense, the two processes work in opposite directions, but within the same region of the brain. Interestingly, some patients with chronic neuropathic pain report relief simply from imagining movement, which suggests that modulating the motor cortex — in either direction — can help alleviate pain.
It’s telling that the different technologies used to reduce pain, such as transcranial magnetic stimulation and transcranial direct current stimulation, all target the motor cortex rather than the sensory cortex, and largely the same location within it. It’s almost as if there’s a single hub from which modulation influences pain more broadly.
Identifying neurological markers that support recovery is central to your work. What have these markers taught you so far about which patients are likely to respond to neuromodulation-based therapies?
That’s the million-dollar question — if these markers could tell me that, I’d be delighted. What they can tell me is, for instance, whether someone is likely to develop pain in the future, or how strong the activation of corticospinal pathways is. But prediction — who will respond to therapy — is much harder.
There are really two separate issues here. Predicting who will respond to movement restoration based on EEG is extremely difficult in spinal cord injury; with stroke patients there’s more to go on, but not in this population. Predicting who will respond to neuromodulation and neurofeedback is a different matter: there is some evidence that people with higher baseline activity in the alpha frequency range tend to be better at controlling their brain waves. That said, I wouldn’t call this a solid finding — it’s fairly speculative, and it isn’t specific to pain; it relates to neurofeedback and neuromodulation more generally, and to the ability to modulate brain waves overall.
As you mentioned, brain-computer interfaces rely on the patient’s own neural signals to drive the therapy. How important is this active, patient-generated component for the effectiveness of the outcomes?
It’s very important. If the patient isn’t actively engaged, essentially nothing happens, and the therapy falls flat. If someone is simply standing there while we record their brain activity, but they’re not actually in control, they’re not modulating the brain networks associated with either motor control or pain. Their ability to control the system is, in effect, their ability to modulate the networks responsible for rehabilitation or pain reduction. So that voluntary engagement is really the key.
Looking ahead, what do you see as the biggest translational challenge in moving closed-loop BCI-based neuromodulation from the lab into everyday clinical rehabilitation?
Obviously, cost — and closely related to that, reliability, together with the fact that you need a medically approved device, which brings you straight back to cost again.
I tested this question on two different audiences. Last year I gave a talk at a neurofeedback workshop in Glasgow, to an audience of around sixty people, mostly engineers. A few months later, I gave a similar presentation at the British and European Pain Society meetings, to an audience of medical doctors. I asked both groups what they saw as the biggest hurdle.
The engineers — and I say this as an engineer myself — focused on cost: the technology is too expensive, patients can’t buy it out of pocket, and so on. Their view was that unless we can produce something cheap that still delivers precise signals, we won’t be able to translate it. The doctors, on the other hand, weren’t thinking about price nearly as much; they were thinking about safety. Their concern was that we need devices with a medical CE mark, and that certification process is extremely expensive.
So, you end up needing to reconcile two things pulling in opposite directions: cheap, and certified-safe. That raises a real question — do we always need the level of medical certification required in Europe? In Australia and New Zealand, for instance, it isn’t required for device-controlled trials. If certification isn’t needed for larger trials, only for eventual commercialization, there’s far more room to test the technology and move it towards translation. Here in the EU and UK, the regulatory requirements are much tougher, and in my view that’s the single biggest hurdle to translating neurotechnology — because cost, time, and the technical resources needed for testing all end up working against each other.
Neuromodulation technologies are increasingly being explored not only for walking but also for bladder function after spinal cord injury. What potential do you see in this area, and how does it relate to the neural mechanisms you study for motor recovery?
I’m aware of that work, and I’d like to see more results. I don’t think it’s impossible, but bladder control isn’t governed the same way — it’s regulated by the pontine micturition center, which sits lower down, below the cortex, so any control you’d achieve there is indirect rather than direct.
That said, I have seen hints that it could work. A few of the people I’ve treated for neuropathic pain reported, as an incidental finding, that their bladder control had improved. I don’t know what proportion of people would respond, since the control isn’t direct, but it’s certainly worth exploring further. There may be some deeper brain structure involved that we can’t easily access with EEG, but that we could reach with a different method — neuromodulation via ultrasound, for instance, rather than voluntary control.
So yes, I think neuromodulation has potential here — whether it acts through the motor cortex or through a deeper structure is a separate question. If it does involve a deeper structure, it wouldn’t need to rely on voluntary modulation at all; it could be achieved with a different technology altogether. As far as I know, ultrasound is probably the cheapest and most efficient option currently available for modulating brain regions that aren’t accessible at the surface.
And does the same potential apply to bowel control?
I’m not entirely sure — bowel control is a different matter. The neurophysiology involved is different: what I said applies to the bladder, but the bowel has its own intrinsic nervous system (the enteric nervous system), so I can’t say with confidence that the same approach would work there too.
