Artificial Intelligence (AI) has permeated virtually every industry of the world, accelerating mankind’s technological capabilities at an almost unthinkable pace. And while AI can navigate your next road trip or impressively predict what kind of paper towels you will likely buy online, the awesome power of this technology can do so much more—even in the incredibly complex field of neurosurgery. I know firsthand how the transformative capabilities of AI and “deep learning” can augment our work, because I have employed these technologies in numerous studies. However, I’m also aware of their limitations, which we can remedy only through more widespread use of the technology in each field.

A Workhorse for Administrative Tasks

We’ve all received those late-night calls in which a worried nurse needs your eyes on a patient’s new brain scan. You ask them to send it immediately, only to easily, quickly determine that the patient is fine. The nurse did the right thing by reaching out to ensure the patient didn’t need immediate care, but these frequent calls for routine review of patient scans can be safely, economically minimized through careful use of AI.

We go through extensive training to be able to treat incredibly difficult, complex diseases of the brain. Our expertise is best utilized in treating these illnesses, but we often find ourselves doing menial tasks, such as reviewing routine brain scans or X-rays that don’t necessarily need a neurosurgeon’s critical eye. Fortunately, new technological applications currently available to neurosurgeons can review thousands of scans per minute.

For example, at Banner Health, we have begun using the first FDA-cleared computer-aided triage system for the rapid detection and notification of suspected large vessel occlusion strokes: Viz.ai. This technology allows our team to more quickly synchronize care and determine optimal patient treatment when compared with standard care. It saves critical minutes, even hours, in the triage, diagnosis, and treatment of stroke. It not only saves the team considerable time, but dramatically helps our patients.

A Helpful Assistant in the Operating Suite

Recent research has shown the merits of AI in the operating room as well. In a piece published in Frontiers in Oncology, my colleague and I examined the capabilities of deep learning and AI to increase intraoperative precision and tailor surgery for malignant invasive brain tumors using confocal laser endomicroscopy (CLE).

CLE images can help diagnose brain tumors. However, sifting through these numerous images is a time-consuming task. AI and deep learning can assist neurosurgeons and pathologists by reviewing these scans (roughly 10 per second) to flag only those that need further review by a neurosurgeon. This imaging analysis paired with pathology—and the surgeon’s own observations—allows us to diagnose and treat a patient far more quickly and accurately.

An Objective Perspective

While AI has proven itself useful in both administrative and technical tasks, it will never replace the complex and nuanced thinking that an experienced neurologic practitioner can provide. But the machine’s very lack of complexity has important benefits, because we humans are fallible.

We try to do our best for our patients, but we can easily fall prey to confirmation bias, a common danger within the medical field. Our biases can often lead to misdiagnoses and incorrect treatments. Clinicians must constantly be aware of these biases to fight them, but AI will always remain objective. This objectivity makes it a valuable tool in ensuring that the decisions we make are primarily based on logic and science, not beliefs or theories.

Significantly, AI also has the benefit of being able to apply its attention more broadly and more frequently to problems than even the most dedicated clinician. By applying this breadth and rapidity of focus, identifying the most critical problems and filtering out the noise, AI can help clinicians focus their attention and expertise.

AI and deep learning can offer a wealth of benefits to neurologists. From reducing time spent on administrative tasks to helping improve intraoperative precision in diagnosis and treatment of malignant brain tumors, AI is advancing the field and, more importantly, patient care. But for AI and deep learning to become even more reliable, more neurosurgeons need to use these valuable tools. The more clinicians use AI, the more these technologies will learn and improve, which in turn will help us do our jobs better. I encourage more neurosurgeons to embrace AI in the hospital, clinic, and operating suite, using these powerful innovations to better serve our patients, improve function, and save more lives.

Author