AI Is “Rewiring” Neurosurgery Through Clinical Decision Support

Considering the critical need for millimeter-level precision in neurosurgery, where any deviation could result in irreversible consequences, emerging applications of artificial intelligence (AI) may help surgeons optimize and refine accuracy beyond manual capacity.

While AI implementation in neurosurgery is in the early stages of revealing future potential, recent advancements in diagnostic imaging, surgical planning, and intraoperative navigation are enabling clinicians to perform individualized procedures that may improve outcomes. Current literature reviews suggest measured optimism and emphasize the need for more high-quality clinical validation of these advancements.

Personalized AI Guides Spine Treatment Pathways

While the integration of AI into spine surgery has yet to reach transformative levels, its footprint continues to expand rapidly.¹

Beginning with preoperative planning, AI is transforming what was traditionally a time-consuming, manual process into an automated, high-precision measurement of the shape and structure of vertebrae, discs, and the spinal canal obtained from dual-energy x-ray absorptiometry (DEXA) images and computed tomography (CT) scans.

“AI is being used in the diagnostic workup, which includes the interpretation of imaging studies,” said Michael G. Fehlings, MD, PhD, FACS, vice chair of research in the Department of Surgery at the University of Toronto in Canada, and a neurosurgeon at Toronto Western Hospital, University Health Network. “AI algorithms can analyze images for issues such as spine morphometry, alignment, and deformity, and quantify these findings.”

These algorithms can provide automated measurements, such as the Cobb angle, which assist surgeons in identifying conditions such as scoliosis, disc degeneration, and stenosis with expert-level precision that can match or exceed human proficiency.²

“Another area where AI is starting to be used is in the assessment of the heterogeneity of patients and personalizing the management of the patient,” added Dr. Fehlings.

In spine care, heterogeneity describes the unintentional variation in patient selection regarding treatment and surgical approaches, which can complicate standardization and decision-making.

To help mitigate variations in care, researchers at the Cleveland Clinic Center for Spine Health in Ohio designed an AI platform that analyzes a comprehensive set of patient data, including lab results and imaging, from more than 55,000 surgeries performed between 2007 and 2022.³

In a study of 3,000 patients who underwent lumbar laminectomy at the Cleveland Clinic, more than half achieved meaningful clinical improvement. Modeling suggested the success rate could exceed 75% with AI-guided decision-making and potential cost savings of up to $25,000 per case.⁴

By using AI to individualize care planning, clinicians can reduce clinical heterogeneity and improve patient outcomes.

AI-Driven Intraoperative Guidance

“In terms of the surgery itself, AI is currently being coupled with a few technologies, including robotics, which is really in its infancy in the area of spine surgery,” explained Dr. Fehlings. “The main use of robotics right now is to assist with the placement of implants such as pedicle screws, although currently less than 5% of pedicle screws in North America are placed with robotics.”

Composed of titanium or cobalt chrome, pedicle screws are used in spinal fusion surgery and function as internal stabilizers while vertebrae heal. Connected by fusion rods to prevent movement, these implants help maintain alignment and typically serve as a treatment pathway for various spinal disorders.

By analyzing patient data and automating specific steps, robotic-assisted screw placement uses AI to improve preoperative planning and optimize real-time navigation that may not be available via approaches that rely solely on manual, freehand, or fluoroscopy-guided techniques.

Traditional screw placement methods have been shown to result in misplacement rates ranging from 30% in the lumbar spine to 55% in the thoracic spine, while at least one study demonstrated a 90% success rate with robotic-assisted pedicle screw placement.⁵

However, while robotic-assisted screw placement offers elevated precision, intraoperative conversion to manual techniques occurs in 7% to 17% of cases examining novice adopters of this technology.⁶ 

“More research is required to validate the use of robotically assisted screw placement,” cautioned Dr. Fehlings. “Right now, this is an emerging technology, and one that has not been validated with multiple, high-quality prospective research studies. Currently, there is a lot of focus on technology development, rather than conducting rigorous studies that examine clinical outcomes.”

AI technology also has a role in creating personalized fusion rods that connect the pedicle screws by analyzing CT and magnetic resonance imaging scans to generate 3D-modeled, customized hardware that aligns with a specific patient’s bone anatomy.

“Currently, the surgeon will take a straight rod and using their eyes, hands, and skills will bend the rod to match the alignment of the spine,” said Dr. Fehlings. “When the surgeon bends the rods intraoperatively, there is potential to weaken the rod during the bending, so it might be more susceptible to breakage. AI-generated, pre-bent rods could reduce revision rates and reduce the time that’s spent in surgery.”

Another AI-driven enhancement to spine surgery is augmented reality (AR), which overlays patient-specific anatomical data onto the surgical field. AR in this context is intended to enhance tasks, including improving pedicle screw placement accuracy and reducing exposure to radiation.

In a review of studies published between January 2004 and May 2025, AR-assisted pedicle screw placement accuracy rates ranged from 93% to 100% with a reduction in fluoroscopy time and improved surgeon ergonomics.⁷ In the article, researchers noted that broader adoption “remains limited by cost, integration challenges, and a lack of large-scale, multicenter trials,” even as this technology shows “early promise in enhancing precision and efficacy.”

The xvision Spine System is reportedly the first AR guidance system to receive clearance from the US Food and Drug Administration, although several other companies offer AR, mixed reality, or advanced image-guided navigation systems tailored for spine surgery.^8,9 The xvision Spine System allows surgeons to “visualize a patient’s 3D spinal anatomy directly on their retina, functioning like ‘x-ray vision’ to navigate instruments and implants during surgery.”⁸

Learning Curve for AI Integration into Spine Surgery

A review of published studies examining the learning curve for surgeons using AI-enabled robotics in spine surgery suggests that 20–30 cases are typically required to attain proficiency, although this number could vary widely depending on a surgeon’s skill and case complexity.¹⁰ Proficiency with this approach could include a number of factors such as demonstrated mastery of preoperative planning tools, robotic navigation, and the ability to interpret AI-driven data.

“I do not see AI replacing spine surgeons or neurosurgeons at any time in the near or foreseeable future,” said Dr. Fehlings. “Over the next 10–20 years, robotics and AI will become part of our day-to-day life, in terms of how we work up our patients and the way we apply the techniques intraoperatively.”

Before this transformation can occur in neurosurgery, clinicians need to cultivate a culture that fully embraces and supports upskilling.

“My advice is that when a wave is coming, it’s much better to surf the wave than to fight the wave,” said Peter Nakaji, MD, FACS, a surgeon specializing in complex cranial neurosurgery, including cerebrovascular surgery and brain tumor treatment. Dr. Nakaji is the founder of Scottsdale Neurosurgery Specialists in Arizona and executive director of the Bob Bové Neuroscience Institute at HonorHealth Scottsdale Osborn Medical Center.

“My father was a general surgeon, and at one point in his career, he watched the laparoscopic revolution unfold, and as a result, he adapted to be able to stay within the field of surgery. The difference today is that previously, those types of changes happened generationally, one or two times in your entire career, while I’ve seen maybe three or four big changes during my career. The key here is not simply getting better, but being very good at getting better. Becoming nimble at learning and adapting to new tools is going to be critical,” he shared.

AI Support for High-Stakes Neurosurgical Decisions

The use of AI-driven technology in brain surgery (and other specialties) often has been described as a “second pair of eyes,” providing surgeons with real-time analysis of the operative field, enabling them to identify tumor margins or anatomical abnormalities that may be invisible to the human eye.

“When I first started in my career, we really just went in with the scan, we committed the image to our own brain, and we did our best to see the tumor,” said Dr. Nakaji. “Some surgeons were really good at seeing the tumor, even where the edge of a tumor just looks like butter melted into margarine. Now, there is a lot more information available to the surgeon. For instance, the image guidance systems can start to detect deformation, and so, even though the scans show one thing, the system can alert you to changes in real time. AI will not be a matter of having another pair of eyes—it’s like having a suite of eyes, which is critical, because there are multiple things that the surgeon must keep track of in the OR.”

For example, surgeons must take into account what they see through their own eyes and whatever is showing via imaging guidance, but they also could have intraoperative neurophysiological monitoring, which features electrophysiological signal feedback to alert them when they are getting close to a structure. By integrating AI into real-time monitoring, these existing feedback systems may further enhance situational awareness.

“You just can’t watch all of the inputs all the time, so we’re always flipping our attention back and forth in order to do that,” said Dr. Nakaji. “I think AI is going to actually provide you much better information than you can get from your own mind because you can’t possibly focus on everything at once. The promise of this capability—that it can provide the surgeon with not just a second set of eyes, but with every eye as it monitors every system—that’s the compelling piece.”

Is AI’s Promise Curbed by Data Deficit?

While AI provides neurosurgeons with real-time intraoperative support as an “always on” surveillance partner through immediate guidance prompting, this technology is constrained by the current dearth of high-quality, standardized metadata, which these platforms rely on to provide reliable information. In this context, data refer to the core clinical information used to diagnose and treat a patient, while metadata offer contextual and descriptive information (in other words, data about the data).¹¹

“I think getting pre-consent from the patient for the AI to capture metadata would be an important step because AI is only as good as the data it is trained on,” said Dr. Nakaji. “Metadata could describe how a procedure actually went, including the surgeon’s impression of the procedure, and everyone would share that data. Right now, those of us in academic or institute models often participate in sharing this information, but this is not the case in community settings largely due to HIPAA regulations.”

If data sharing were amplified, surgeons would conceivably benefit from the experiences of every other surgeon that has used the system, including lessons learned from missteps.

“Recently, I spent all day taking out a giant acoustic neuroma from somebody’s skull base,” he said. “There are three approaches to that tumor, and in fact, there are even more sub-approaches within those techniques. However, surgeons often specialize in only one or two techniques that they are most comfortable with, sometimes in preference to other valid options.” 

Surgeons should be open to all relevant approaches with the goal of providing customized surgical care to each individual patient.

“We talk about standardization, for example in joint replacement surgery, and for those cases, the more standardized, probably the better,” added Dr. Nakaji. “But for brain surgery, you cannot do the same thing every time. The question of how to customize these cases is driven by experience—and not everyone can have the necessary experience. But you can use the AI to help you to function like you have it.”

The amount of training necessary to become proficient in AI-augmented brain surgery, with neurosurgeons employing robotic-assisted or virtual reality technology, for example, is difficult to ascertain at this point for many reasons, including high variability in how quickly clinicians master these associated skills.

“We haven’t determined the learning curve for this, but I can tell you that while surgeons mostly learn on their feet in the OR, I don’t think this will continue in the future because there will be simulation models similar to what airplane pilots use during their training,” said Dr. Nakaji. “With AI simulation, the system will watch the trainee and say, ‘You’re not getting this part, and I think I see why.’ Or ‘You don’t know the anatomy well enough,’ or ‘There’s a basic hand-eye coordination issue,’ and then the trainee would work to improve on these issues. We can’t afford the cost to have a top surgeon coach you every day, but AI could get you at least part of the way.”

By providing real-life, complex operative scenarios in a risk-free setting, AI-enhanced simulation may help improve skill acquisition and accuracy.

“I think there’s a beautiful new era of surgery coming where these systems will be in place and surgeons will be much better equipped. They’ll be like Iron Man,” said Dr. Nakaji, referring to the comic book superhero whose suit of armor provides enhanced capabilities. “The surgeon will feel like they’re in an environment that gives them superpowers.”

Tony Peregrin is the Managing Editor of Special Projects in the ACS Division of Integrated Communications in Chicago, IL.

References

1. Muelbauer EJ, Alvi MA, Kennedy DJ, Fehlings MG. The future is now: How AI is reshaping spine care. N Am Spine Soc J. 2025;24:100825.    
2. Lee S, Joon-Young J, Mahatthanatrakul A, Kim JS. Artificial intelligence in spinal imaging and patient care: A review of recent advances. Neurospine. 2024; 21(2):474-486.
3. Habboub G, Berven S, Ames C, Peterson T, et al. Real-world implementation of artificial intelligence/machine learning for managing surgical spine patients at 2 academic health care systems. Int J Spine Surg. 2023;17(S1):S11-S17.
4. Cleveland Clinic. Can artificial intelligence point the way to better spine care? March 21, 2023. Available at: https://consultqd.clevelandclinic.org/can-artificial-intelligence-point-the-way-to-better-spine-care. Accessed April 13, 2026.
5. Gatam L, Phedy P, Husin S, Mahadhipta H, et al. Robotic pedicle screw placement for minimal invasive thoracolumbar spine surgery: A technical note. Front Surg. 2025;11:1495251
6. Hu X, Lieberman IH. What is the learning curve for robotic-assisted pedicle screw placement in spine surgery? Clin Orthop Relat Res. 2014;472(6):1839-1844.
7. Nadeem-Tariq A, Kazemeini S, Kaur P, Dang G, et al. Augmented reality in spine surgery: A narrative review of clinical accuracy, workflow efficiency, and barriers to adoption. Cureus. 2025;17(6):e86803.
8. Augmedics. X-vision Spine System. Available at: https://augmedics.com/xvision/. Accessed April 13, 2026.
9. Sandberg J. First augmented reality spine surgery using FDA-cleared Augmedics vision Spine System completed in the US. Ortho Spine News. June 11, 2020. Available at: https://orthospinenews.com/2020/06/11/first-augmented-reality-spine-surgery-using-fda-cleared-augmedics-xvision-spine-system-completed-in-u-s/. Accessed April 13, 2026.
10. Pennington Z, Judy BF, Zakaria HM, Lakomkin N, et al. Learning curves in robot-assisted spine surgery: A systematic review and proposal of application to residency curricula. Neurosurg Focus. 2022;52(1):E3.
11. Sugiyama T, Sugimori H, Tang M, Fujimura M. Artificial intelligence for patient safety and surgical education in neurosurgery. JMA J. 2025;8(1):76-85.