Over the past few years, endovascular robotics has re-emerged as a notable ‘hot topic’ in the neurointerventional space. While this revival appears to have been partly driven by new industry players working to develop the latest generation of these technologies, several leading physicians are also advocating the advantages of robotics in the treatment of various neurovascular conditions. However, as with any novel introduction on which high-level evidence remains limited, endovascular robotic systems have their vocal detractors. Here, NeuroNews speaks to a neurointerventionist from each of these camps to get their unfiltered perspectives on one of the most contentious talking points currently facing the field.
Subsequent to its 2019 acquisition of Corindus’s CorPath GRX robot, Siemens Healthineers discontinued the system’s usage for cardiology procedures in May 2023, choosing instead to focus more intently on neurovascular indications. Last year, six-month data on more than 100 brain aneurysm patients treated endovascularly using the CorPath GRX system were shared for the first time, demonstrating positive technical and clinical outcomes. Siemens Healthineers also announced shortly thereafter a collaboration with Stryker to advance robotic technologies capable of performing ischaemic and haemorrhagic stroke procedures alike.
Along a similar timeline, several smaller companies—including but not limited to Nanoflex, Microbot, Remedy, Sentante, Stereotaxis and XCath—have publicised their own milestones relating to neurointerventional robotics.
In October 2025, Remedy debuted its artificial intelligence (AI)-powered N1 system and announced what it described as the world’s first entirely remotely operated, in-human endovascular interventions requiring no bedside interventionist. The following month saw XCath disclose the successful in-human use of its system in three patients with complex brain aneurysms, dubbing this “the second time in history” that a surgical robot had been used in an intracranial neurovascular intervention specifically.
Earlier this year, in March, XCath also claimed to have completed the world’s first telerobotic stroke procedure, with Vitor Mendes Pereira (University of Toronto, Toronto, Canada) leading a remote mechanical thrombectomy from Clinica Norte in Santiago, Panama—more than 200km away from a patient located in Panama City. The Iris system deployed during this procedure went on to gain a US Food and Drug Administration (FDA) Breakthrough Device designation for remote, robotic-assisted thrombectomy in acute ischaemic stroke in September 2026.
Having gained its own US FDA Breakthrough Device designation last September—followed by CE-mark approval in May this year—Sentante announced the completion of 11 full end-to-end thrombectomies performed transatlantically on perfused human cadaver models using its endovascular robotic platform in October 2025. Weeks later, the same platform was utilised in living subjects for the first time, with animal models being operated on remotely by physicians located in the UK, USA and Lithuania.
Iris Grunwald (University of Dundee, Dundee, UK), who led the UK-based procedure, subsequently highlighted the significance of this particular system being device-agnostic, as well as describing the haptic feedback it provides as “exceptional”.
“You feel exactly what the device is doing, even though you’re operating it remotely,” she stated at the time of the announcement. “The subtle, tactile cues give you confidence in every movement. It’s almost like having your hands directly on the catheter, but with even greater stability.”
Technological advances
“If you’d asked me 18 months ago, I would’ve said [these systems] are not there yet—but they’ve made a very big leap forward,” Grunwald comments, speaking to NeuroNews to reflect on these recent milestones and the broader landscape of neurovascular robotics. “The first CorPath robot was essentially useless. It took a long time to load the devices, you needed assistance to get all the way up [into the neurovasculature], it could only be used to do the final part of the procedure—capturing the thrombus—and the feedback and torque-ability just weren’t there.
“So, previously, I would have said there is too much of a time delay and it would be better to train other specialties to do thrombectomy [to make these procedures more accessible]. But this robot is very intuitive, it has a very easy setup, and the latency is about the same as the blink of an eye—a delay of 36–136 milliseconds has been reported—which is not going to negatively impact the intervention. At first, I thought this was just a gimmick but, now, having seen and felt what this robot can do, I am completely converted.”
Grunwald’s view is that the capacity for a robotic system to deliver haptic feedback to the operator is a vital consideration—particularly in the neurointerventional sphere.
“The vessels in the brain are really tiny, and there are also lots of perforators,” she adds. “If a vessel is blocked, we cannot see what is behind it. That means we often advance our wires and tubes relying on haptic feedback, and, if there is resistance, we don’t continue to push. That’s what prevents us from advancing larger catheters and causing perforations. As such, I do not think any robotic device without haptic feedback or torque-ability will survive.”
Expanding on the need for these systems to also be device-agnostic—meaning existing, approved wires, catheters and stents can be selected at the operator’s discretion—Grunwald suggests this is also a prerequisite for any robotic technology looking to stand the test of time.
“Many of the companies are developing specific devices solely for their robots,” she continues. “That’s not going to work, because you’ll have devices that will expire, each operator is familiar with and has preferences for different devices, and some devices may be better than others in specific situations. You have to be able to use any device you have on the shelf and that you are comfortable with in order to reduce the learning curve.”
Reflecting on the landmark procedures she co-led late last year from Dundee, Grunwald also recalls a “nightmare scenario” whereby, during the operation, there was a complete power outage in Paris, France, where the porcine model itself was located.
“Normally, if a hospital has a power outage, the angio suite will go onto a backup generator that gives you an extra 20 minutes to complete a rescue strategy,” she says. “But, because this was an animal lab, they didn’t have one, so they needed to hand ventilate the pig and they had to use torches in order to see. The only thing that was still working was the robot, because it runs on mobile data networks and also has built-in batteries. That was quite reassuring, because I was able to finish the intervention without switching to a manual procedure, and I think it’s important to see what the robot does in those situations.”
In spite of the various achievements announced over the past couple of years, there are still those within the neurointerventional space who feel—as Grunwald once did—that the re-emergence of robotics is little more than a fad driven by industry.
One of Christophe Cognard’s (Toulouse University Hospital, Toulouse, France) primary reservations centres on the fact that robots add minimal benefit relative to what can be achieved by physicians themselves.
“I don’t see the value of having a machine to do something that a human can do with their hands, quite easily and efficiently, after two years of training,” he comments. “And, for me, the complexity of neuroendovascular treatment is in the brain [of the physician] rather than the hands. Proceduralists should not become too proud of what they can do with their hands. You have to perfectly understand the patient’s disease, and then make decisions about what to do and how you will do it. The majority of our job is intellectual, not manual.
“People also consider that these things could be done better by a robot than by a human hand, but I don’t think that’s true at all.”
What Cognard alludes to here is the notion that robotic systems may one day be able to actively enhance the precision of neurovascular procedures, as opposed to merely facilitating operators in performing these treatments from a remote location. While she believes there is potential on this front, Grunwald corroborates the view that optimising existing techniques is not likely to be the first port of call for robotics.
“In the future, there may be submillimetre precision and movements that can even counteract tremor in the operator’s hand,” she says. “Robotics may even be able to automate the more repetitive steps within a procedure, and magnetic steerability could be useful in leading the wire directly into an aneurysm—but I don’t think we’re there yet. They have the potential to [enhance] what human operators are capable of, but I don’t think there is the imminent need for this, and the technology is not far enough along right now.”
Is there a genuine need?
In what context, then, can today’s robots truly influence patient care for the better and in a meaningful way?
In addition to the technical progress that has been made in recent years, Grunwald believes an increased “pressure” on providing equitable access to mechanical thrombectomy has also played a significant role in rekindling interest in these systems. Here, she warns against development that moves in “the wrong direction” relative to what is needed by physicians and their patients.
“At the moment, I think everyone is trying to build an all-singing, all-dancing robot with many different arms [and capabilities],” Grunwald explains. “The question is, where is the true indication for robotics? For me, the biggest indication is in thrombectomy, particularly in centres that have the least money and the fewest resources. Are they all going to have these fantastic robotic systems that require a high level of maintenance and probably use specific devices developed solely for that robot? I don’t think so.”
It’s a matter of politics, and of business—and people, including physicians, want to be part of that business because they will make a lot of money.
Christophe Cognard
Grunwald’s solution to this hinges on the idea of two different and fairly distinct types of robotic systems, with the first variant being one that can be installed quickly and easily at any hospital, is relatively cheap, and enables manoeuvrability of a maximum of three or four catheters in order to facilitate thrombectomy.
“It’s not just about being able to instal the robot,” she continues. “We need the procedures to be faster than transporting the patient [to a thrombectomy-capable centre]. That’s the important part, and it requires a stroke network. I think, in five years, we will have comprehensive, telerobotic stroke networks, because development is moving so quickly. But there cannot be a huge learning curve for interventionists to use these robotic systems, and they need to be simple, cheap and practical.”
Given that every additional six minutes following a stroke equates to a 1% reduction in the likelihood of a good functional outcome, the most obvious and pressing use case for remote, robotics-led interventions appears to be in more rural or sparsely populated areas where patients may face delays of several hours before they can be transferred to a thrombectomy-capable hospital. A more efficient and ultimately affordable system may therefore hold the key to expanding the reach of stroke interventions beyond the walls of dedicated centres, and delivering 24/7 thrombectomy services on a more consistent basis.
Meanwhile, in Grunwald’s view, “all-singing, all-dancing” robots of the more long-term future—incorporating greater imaging capabilities, magnetic steerability and various forms of AI assistance—are likely to be deployed at larger hospitals as part of an interventional neuroradiologist’s day-to-day practice, including for more complex cases. This eventuality, however, represents a less immediately obvious clinical need or health-economics justification, with the primary advantages being reduced radiation and improved comfort for the operator.
“I think we need two robots, and I’m hoping that these companies will not just try to compete with each other over who has the biggest and best system that can do the most, because then everything will become much more expensive,” Grunwald adds. “I’m hopeful they can give us a simple, robust solution for thrombectomy procedures.”
Recent landmark procedures like XCath utilising its robotic system for an intracranial aneurysm procedure may not be representative of the most pressing clinical indication as of today, but Grunwald believes such milestones do serve as a valid proof of concept.
“If we have a perforation during a thrombectomy, the only way to solve this is via endovascular rescue techniques—similarly to aneurysm treatments [involving coils or balloons]—to close that perforated vessel off,” she explains. “So, a robot needs to allow deployment of a coil for emergency rescue procedures, and showing that this is possible was a good idea. But, at the moment, I don’t believe there is a clinical need to cover everything and for aneurysms to be treated robotically.”
Hurdles to overcome
Even in a world where the theoretical benefits of robotic systems are accepted and embraced—which is no guarantee—there are several practical considerations that are yet to be fully addressed.
Cognard feels that, among those championing these technologies, there is a failure to recognise that neurointerventional care involves a full team of specialists, each with their own essential role to play in maximising the chances of a positive outcome. Inside the angio suite, the surgeon is typically joined at the operating table by an assistant and a nurse, while multiple technicians as well as a neuroanaesthesiologist and anaesthesiology nurse may also be present.
“So, there could be seven or eight people in the room, and all of them are highly important,” he says. “The fact all of these team members are very well trained is the most important point in making the procedure safe and efficient—and the robot will only replace one of them.”
Cognard also highlights the need for a team member to be present onsite to prevent air bubbles and clots accumulating inside each catheter.
“People might consider that, with the robot, I can be in Toulouse operating on a patient on the other side of the world, but it’s not true, because the whole team has to be very well trained. It is important that the team members are used to working together and that it is a collective effort,” he continues. “People are also saying the robot will be used in the middle of Australia, the north of Sweden, or parts of the USA where there is less activity—but, if there is only one thrombectomy per month in these places, none of the team will be performing well.”
There is a fear among many that, if operators and other clinicians involved in neurovascular procedures become heavily reliant on robotic systems, there could be a dropoff in their experience and overall skill levels, which becomes problematic when a manual procedure is required. Challenges around skill retention may also be encountered regarding the actual usage of robotic systems if operators are only performing these types of interventions on a very infrequent basis.
If you’d asked me 18 months ago, I would’ve said [these systems] are not there yet—but they’ve made a very big leap forward.
Iris Grunwald
Another elephant in the room is the “incredible cost”—in Cognard’s words—of deploying robotic systems. As he states, any neurointerventional procedure is unavoidably expensive, with this being a key reason for uptake of thrombectomy—despite its proven benefits—remaining very limited in poorer regions.
“The cost of these robots is enormous, and something that increases the cost of these treatments in an unreasonable fashion is not acceptable,” Cognard notes, before also questioning the sustainability of robotic technologies, owing to the large quantities of disposable plastics they use, as well as the “tremendous amounts” of both energy and data they require.
“It’s a very big business,” he posits. “It’s very good for the companies, and it’s very good for the doctors at congresses saying, ‘I’ve done a case with a robot’. People are excited about these technological developments, but there’s no need for it. It’s a lot of money and energy for a drop in the quality of patient care.
“It does not solve anything in terms of lack of organisation. In countries that don’t have enough angio suites, or the necessary imaging systems, or stroke units or INR [interventional neuroradiology] teams, robotics will not solve the problem at all, because you will need to have people on the ground, everywhere, in every country. It’s about the entire team, not just the operator, and they will need to be trained to do remote cases just as they would with local cases. Endovascular robotics is a dream created by a fascination with technology but, in reality, it’s completely useless.”
While she reiterates that reducing the upfront costs of robotic systems must remain a priority, Grunwald does not share Cognard’s concerns regarding longer-term cost barriers.
“We know how much more expensive a patient becomes [for the healthcare system] with each category of mRS [modified Rankin scale], so the cost-effectiveness in stroke is immediately clear and will be shown very easily,” she states.
Grunwald also feels that there is a “huge political element” to these debates. In her opinion, underlying the “scepticism and defensiveness” of many robotic detractors is the fact that they are “afraid they will lose their jobs”—perhaps not today, or even within the next few years, but potentially further down the line if robotics-driven automation of endovascular procedures becomes more commonplace. She notes, however, that these reservations are misplaced when it comes to robotic-assisted procedures.
“You’re still being treated by an interventionist,” Grunwald explains. “It’s not the robot that treats you—which I think is often miscommunicated. You’re still going to have the same hand movements, the same decision-making and expertise, and the same choices over devices and access.
“I would personally feel comfortable being treated [robotically] by anyone with neurointerventional experience, and that’s always how I judge these things. I don’t think it will replace neurointerventionists; we, the operators, are still in charge and doing the procedure, and I think acceptance will work very well once people have had these systems in their hands and experienced how lifelike they are.”
What does the future hold?
Despite their opposing stances on robotics, Cognard and Grunwald have both advocated the training of non-neurointerventional specialists as a means for broadening access to thrombectomy—via formal European Society of Cardiology (ESC) guidance on the interdisciplinary management of ischaemic stroke, and the 2023 World Federation for Interventional Stroke Treatment (WIST) guidelines, respectively.
“The [ESC] guidelines state that cardiologists can become neurointerventionists as soon as they’ve done the requisite training, which takes two years—but, by law, every cardiologist can do a thrombectomy,” Cognard comments. “It depends on the healthcare system in each country. We need to have more centres, operators, angio suites and stroke units. It’s a matter of politics, and a willingness to develop healthcare systems and stroke networks, to buy angiography machines, and to train enough people.”
While Grunwald remains a believer in cross-specialty training, recent technological advances have convinced her that robotics also have a crucial role to play.
“We still need some basic infrastructure,” she states, acknowledging one of Cognard’s prevailing concerns. “We need an angio suite, we need someone capable of doing arterial access—at least at this point—we need monitoring capabilities, we need someone to sedate our patient or administer general anaesthesia, and we need someone to manage their postoperative care as needed. We need the local site to have the equipment [and personnel] to deal with emergencies as well as certain other aspects too.”
Additionally, Grunwald notes that successfully establishing robotic stroke networks will be key to ensuring skill retention and guaranteeing standards of care remain consistent.
“The robot is no longer the challenge,” she adds. “I don’t think they’re ready for everything else but, for thrombectomy, they are ready. The challenge lies in setting up networks and being able to immediately communicate with each other.”
Heading into 2027, clinical studies evaluating these robotic systems are an imminent inevitability, with Sentante and other companies already targeting more widespread evaluation. In addition to assessing the feasibility of setting up robotic thrombectomy teams, Grunwald believes the focus of these studies should be to analyse the performances of these teams—as opposed to the robots themselves. In her view, the main outcome measures in such studies will likely be comparisons of time to treatment before and after robotic implementation, as well as the safety of the interventions, including any complications and how they were dealt with.
“I often hear people say these sites need to be able to manage any emergency or complication that could occur,” Grunwald comments, “But, here, we’re not talking about treatment at a specialised neuroradiology centre versus treatment somewhere else; we are comparing no treatment, or significantly delayed treatment, versus robotic treatment. That’s also why I believe informed consent won’t be a problem from an ethical perspective, because the options will be not receiving thrombectomy treatment—where we know how bad the outcomes are—or a robotic thrombectomy if you otherwise would not be able to make it to a centre offering this procedure.”
Cognard, on the other hand, does not see a place for robotic systems when he looks to the horizon of neurovascular care.
“It’s a matter of politics, and of business—and people, including physicians, want to be part of that business because they will make a lot of money,” he argues. “We need to stop being fascinated by everything that is novel, and we should not accept a new technique without evidence from several RCTs [randomised controlled trials] showing that using robotics is beneficial in terms of patient care. We should also consider the environmental cost in terms of carbon, electricity, and plastics.
“I would say it will never fully develop, but it is such a big business and it’s good for the companies. So, it could be the case that it becomes routinely used, but it’s purely financial. In reality, do these physicians believe that, in the future, every patient in the world will be treated by a robot? I don’t think so. I think they are biased and they have too much of an interest in the field. I cannot see how people can defend the fact that they will use a robot.”
Ultimately, time—and, with it, further real-world experiences and evidence from clinical studies—will tell whether robotic systems have a legitimate role to play in the future of neurointerventional care, as well as elucidating what that role might be.
