The integration of artificial intelligence and robotics in cerebrovascular surgery has the potential to bring about significant improvement in patient care, surgical outcomes, cost-efficiency, and training of cerebrovascular surgeons. These advancements can revolutionise neurointerventional procedures, address unmet needs in acute stroke care, and shape the future of cerebrovascular surgery globally.
Cerebrovascular surgery encompasses a range of specialised procedures performed by neurosurgeons to address blood vessel conditions affecting the brain. These surgeries can be broadly classified into two categories: open surgeries, which involve direct access and manipulation of cerebral vessels, and endovascular interventions (or neurointervention), which are minimally invasive procedures accessing cerebral vessels through distant sites in the wrist or thigh. With the intricate nature of the skull base and the significant burden on patients associated with open neurosurgery, endovascular interventions have emerged as the preferred initial approach in cerebrovascular surgery. Open surgery is reserved for cases with challenging anatomy or disease processes for endovascular treatments.
Endovascular interventions involve the delicate navigation of wires and catheters through the brain's arteries using continuous X-ray fluoroscopy for real-time catheter positioning. Initially recognized for their effectiveness in treating aneurysms and arteriovenous malformations, these procedures have experienced significant growth in the treatment of acute ischemic stroke with mechanical thrombectomy. Acute ischemic stroke occurs when a blood clot obstructs an artery that supplies the brain, leading to oxygen deprivation and subsequent brain tissue damage. Several clinical trials have unequivocally demonstrated the superiority of mechanical thrombectomy, an endovascular procedure that removes blood clots from arteries, over traditional medical management for certain types of strokes caused by occlusion of major arteries supplying large portions of the brain. With a number-needed-to-treat of 2.6, mechanical thrombectomy carries the highest magnitude of benefit in reducing stroke-related disability among all stroke treatments, and is one of the most effective interventions. Mechanical thrombectomy not only offers improved functional independence for patients but also is cost-effective, resulting in savings of over US$18,000 per patient and more than US$105,000 per qualityadjusted year of life in a clinical outcome study conducted in Netherlands5. Despite the compelling evidence supporting mechanical thrombectomy, a global survey conducted by Mission Thrombectomy 2020+, a global network of experts focused on improving access mechanical thrombectomy, found that the median of 2.8% of eligible stroke patients across 67 countries undergo mechanical thrombectomy, highlighting a significant potential increase in demand for endovascular interventions and the pressing need to expand comprehensive stroke care systems.
Stroke is a leading cause of mortality in China with an estimated 343 per 100,000 person-years. It remains the second-leading cause of death globally, with approximately 12 million new cases annually and an estimated global cost of over US$721 billion. Timely removal of the offending blood clot causing a stroke is critical to achieve favourable patient outcomes. Research suggests that approximately 1.9 million brain cells are lost per minute when a stroke is left untreated. Mechanical thrombectomy, a procedure capable of treating large vessel occlusions responsible for about 30 per cent of acute strokes, is highly effective at preventing stroke-related disability. However, despite the proven benefits, among the 24 Asian countries included in the MT-GLASS study, a median of only 6.4 per cent of eligible patients received mechanical thrombectomy from 2020 to 2021, with a significant disparity between the countries with the highest (Bahrain, 45 per cent) and lowest (Bangladesh, 0.1 per cent) treatment rates among countries with a nonzero access to mechanical thrombectomy. Notably, the two most populous countries, China and India, reported only treating 7.3 per cent and 1.8 per cent of eligible patients, respectively, suggesting a massive patient population in Asia with an unmet need for neurointerventional stroke care. The underlying factors contributing to this care gap are multifaceted, stemming from the urgent nature of stroke treatment, the geographical distribution of potential stroke patients, and the scarcity of specialised centres and surgeons trained to deliver neurointerventional care.
To illustrate, a significant percentage of stroke patients in the United States, especially those in rural areas (over 50 per cent of all stroke patients and 75 per cent of rural patients), with strokes caused by large vessel occlusions do not have immediate access to centres equipped for mechanical thrombectomy. Training additional surgeons and establishing more stroke centres seem like straightforward approaches, but they come with their own set of challenges. The training for neurointerventional surgeons is rigorous and lengthy, requiring fellowship-level training and exposure to a significantly volume of cases. Simply increasing the number of stroke intervention centres and staffing them with physicians may not guarantee the dissemination of high-quality care. Studies have shown that centres with higher case volumes consistently demonstrate better patient outcomes, suggesting that concentration of expertise leads to improved results. This creates a dilemma: expanding the number and geographic distribution of neurointerventional surgeons may reduce the time to treat, potentially improving outcomes. However, it may also dilute the level of expertise available, compromising patient outcomes. Achieving the ideal scenario of having an expert present everywhere simultaneously is impractical and costly.
Endovascular robotics and artificial intelligence (AI) provide promising solutions to this dilemma. Endovascular robots offer the possibility of remote presence, enabling a distributed network of robots that can extend the “hands” of an expert surgeon to areas where specialised expertise is limited. Coupled with AI, endovascular robotics can further present opportunities for improving the quality and accessibility of cerebrovascular surgery. AI algorithms can assimilate and learn from the collective experiences of numerous neurosurgeons, creating a "super expert" that can continuously improve over time. This “super expert” can serve as an autopilot during endovascular interventions, providing guidance and support to less-experienced surgeons or operating autonomously when needed. The integration of endovascular robotics and AI holds great promise for improving the accessibility, quality, and cost-effectiveness of cerebrovascular surgery.
The integration of AI and robotics in cerebrovascular interventions has the potential to revolutionise healthcare systems by extending life-saving neurointerventional care to a significantly larger number of patients who would otherwise face barriers to accessing specialised centres promptly, which can lead to substantial improvements in patient outcomes, reduced disability, and cost savings.
The combination of AI and robotics in neurointerventional procedures has the potential to enhance safety and accuracy by mitigating the risk of human errors, fatigue, and distraction, particularly during complex and lengthy procedures. Additionally, integrating AI with robotic systems can address the issue of varying skill levels and experience among human operators, thereby improving the standardisation and consistency of neurointerventional procedures. By automating tasks such as catheter navigation and device deployment, self-driving robots can maintain a consistently high level of precision and accuracy throughout the procedure, ultimately improving patient outcomes and reducing complications. This standardisation improves the predictability and reliability of results and promotes the adoption of best practices in the field.
In addition to improving patient outcomes, AI-augmented robotic neurointerventions can increase the efficiency and productivity of healthcare centres. With automated systems performing certain tasks more efficiently, timely, and accurately, potentially leading to faster diagnoses and treatments. This increased efficiency allows medical centres to handle a greater volume of procedures, maximise resource utilisation, and enhance overall productivity. Furthermore, the integration of self-driving robotic systems in neurointerventional care has the potential for significant cost savings. By improving patient outcomes and reducing complications, there can be substantial savings associated with reduced disability and longterm healthcare costs. Additionally, the increased volume of procedures performed per centre, facilitated by the efficiency of robotic systems, can lead to cost savings through economies of scale and resource optimisation.
Overall, the benefits of integrating AI and robotics in cerebrovascular interventions extend beyond improved patient access and outcomes. These advancements enhance the safety, consistency, and efficiency of procedures,
leading to cost savings for the healthcare system. By leveraging advanced technology, healthcare centres can provide high-quality neurointerventional care to a larger patient population, improving lives and transforming healthcare delivery.
Endovascular robots have the potential to significantly enhance the productivity of the neurointerventional physicians by alleviating cognitive and orthopaedic stress, leveraging automation for timesaving precision, and revolutionising training through remote proctoring and automated feedback.
By allowing operators to sit in a radiation-shielded enclosure, robotic systems minimise the occupational risks of ionising radiation and orthopaedic stress, providing a safer working environment for neurointerventional radiologists and neurosurgeons. The ergonomic setup allows the physician to view the fluoroscopy display from a sitting position, just a few feet away, without the burden of lead and sterile gowning. This proximity facilitates natural interaction with computer vision and image processing functionalities, enhancing the physician's ability to interpret and analyse complex anatomy. Moreover, endovascular robots enable superhuman dexterity, with large joystick deflections translated into precise millimeter-scale movements of intravascular devices. Automated force sensing further enhances the precision of procedures, making them less reliant on operators' physical senses and improving overall safety and consistency.
Automation of certain aspects of interventions, such as navigation, can bring significant benefits to neurointerventional procedures by saving time and allowing interventional physicians to focus on higher-level tasks. For example, in mechanical thrombectomy for stroke, navigating through tortuous vessels can be challenging and time-consuming. Similarly, in diagnostic procedures, the primary task of the neurointerventional physician is to safely navigate the catheter to the target vessels. By relieving neurosurgeons or interventional radiologists from the burden of catheter navigation, automated systems can empower technologists and supporting staff to acquire angiography images, freeing up the physician's time for reviewing and interpreting the images. This logistical flexibility allows the physicians to efficiently schedule diagnostic procedures alongside their open surgeries and other interventional cases.
Furthermore, the integration of robotics and AI has the potential to revolutionise the training experience for new neurosurgeons and extend the reach of experienced mentors, addressing the shortage of skilled neurointerventional physicians. Robotic telepresence enables experienced neurointerventional physicians to teach and mentor new physicians remotely, overcoming geographical barriers. Computer vision techniques can further enhance the training experience by automatically assessing trainees' abilities, provide personalised feedback to trainees for improvement, and generate comprehensive recordings of their performance, similar to athletes reviewing game footage, enabling in-depth analysis and facilitating growth and learning. The automation capabilities of robotic systems can reduce the impact of individual skills and enable trainees to attain an expert level of proficiency at an accelerated pace. These strategies not only enhance the training pipeline but also foster the development of a robust neurointerventional workforce.
There are technical and operational challenges that could hinder wide adoption however. First, remote interventions cannot be conducted by a remote surgeon alone. At minimum, trained on-site technologists and an on-site physician capable of establishing the initial vascular access are necessary. Remote interventions are predicated on a stable and fast network connection, and delays between a surgeon’s input commands and catheter motion could impact safety and efficacy. Although rare, remote endovascular interventions may have intra-operative complications, such as bleeding, embolism or swelling, that may require conversion to open neurosurgery. Even in the absence of intra-operative complications, some patients will need to be monitored in neurological intensive care units. In these scenarios, transfer is inevitable, and attempting to save time with a remote intervention may have equivocal or negative impact on these patients’ final outcomes.
The integration of AI and robotics in cerebrovascular surgery heralds a new era of innovation that offers substantial benefits to the healthcare system, neurointerventional physicians, and patients. These technologies have the potential to overcome barriers to patient access, improve outcomes, and enhance cost-efficiency. By empowering physicians with advanced capabilities, ensuring patient safety, improving training experiences, and enabling autonomous technology, these advancements will revolutionise the field of cerebrovascular surgery and deliver exceptional healthcare outcomes worldwide.
