History of neurosurgery: from trepanation to AI

Neurosurgery is one of the oldest surgical practices and one of the youngest specialties. People were drilling holes in skulls more than 7,000 years ago, but modern neurosurgery only took shape in the late 1800s and early 1900s, after anesthesia, antiseptic technique and an understanding of which brain areas control which functions. Today the field is adding a new tool: machine learning, which helps read scans, predict outcomes and guide decisions.

What was the first brain surgery in history?

The first known brain surgery is trepanation: cutting or scraping a hole in the skull. Skulls with deliberate surgical openings have been found from the Neolithic period, some more than 7,000 years old, across Europe, Africa, Asia and the Americas.

Many of these skulls show smooth, healed bone around the opening, which means the patient lived for months or years afterward. Ancient Peru has one of the richest records, with hundreds of trepanned skulls and survival that appears to have improved over time as techniques were refined. Why it was done is still debated. Some cases likely treated head injuries, and others may have been ritual or attempts to relieve headaches or seizures.

Who wrote about brain surgery in the ancient world?

Ancient physicians wrote down practical rules for head injuries. The Hippocratic text On Injuries of the Head, from Greece around the 5th century BCE, describes skull fractures and when trepanation should be done. Later, the Andalusian surgeon Al-Zahrawi (known in Europe as Albucasis) described surgical instruments and the treatment of skull fractures in his encyclopedia of medicine around the year 1000.

For centuries afterward, opening the skull remained rare and dangerous. Infection and pain made anything beyond treating fractures close to impossible.

When was the first successful brain surgery?

Three advances in the 1800s made real brain surgery possible:

• Anesthesia. Ether was first demonstrated publicly for surgery in 1846, so operations no longer had to be fast.
• Antiseptic technique. Joseph Lister's methods, published from 1867, sharply reduced deadly infections.
• Cerebral localization. Physicians such as Paul Broca showed that specific functions live in specific brain areas, so a patient's symptoms could point to where a lesion was.

With these in place, surgeons began removing tumors. In 1879 William Macewen in Glasgow removed a tumor of the brain's covering (a meningioma) located from the patient's symptoms, and the patient survived. In 1884 in London, Rickman Godlee removed a brain tumor that the neurologist Alexander Hughes Bennett had localized from the patient's symptoms. That patient died of infection several weeks later, but the operation showed that localization could guide a surgeon to a tumor inside the brain. In 1887 Victor Horsley removed a tumor compressing the spinal cord, and he went on to become one of the first surgeons to specialize in the nervous system.

Who is the father of neurosurgery?

Harvey Cushing (1869 to 1939) is widely called the father of modern neurosurgery. Working mainly at Johns Hopkins and then Harvard, he made brain surgery survivable on a routine basis. Early brain operations had very high death rates. Cushing brought operative mortality for brain tumors down dramatically through meticulous technique, careful control of bleeding and close monitoring of the patient.

His contributions include:

• Monitoring blood pressure during surgery.
• Bringing electrosurgery (cutting and sealing vessels with electric current) into neurosurgery, working with the physicist William Bovie in the 1920s.
• Describing and classifying brain tumors, including meningiomas, from thousands of operations.
• Describing the hormonal condition now called Cushing's disease, caused by a pituitary tumor.
• Training a generation of surgeons who founded neurosurgery programs around the world.

Victor Horsley in Britain is often named alongside Cushing as a founder of the specialty.

How did imaging change neurosurgery?

Before modern imaging, surgeons had to infer where a lesion was from the neurological exam alone. A series of imaging advances changed that:

• 1918 to 1919: Walter Dandy introduced ventriculography and pneumoencephalography, which used air to outline the brain's fluid spaces on X-rays.
• 1927: Egas Moniz introduced cerebral angiography, making the brain's blood vessels visible.
• 1971: The first CT scan of a patient's brain was performed, based on Godfrey Hounsfield's invention. For the first time, surgeons could see the brain itself in cross-section.
• 1980s: MRI entered clinical use, showing soft tissue in far greater detail without radiation.

Each step let surgeons plan more precisely and operate through smaller openings.

History of neurosurgery timeline

Year	Milestone
More than 7,000 years ago	Trepanation practiced in Neolithic societies
About 5th century BCE	Hippocratic writings on head injuries
About 1000	Al-Zahrawi describes skull fracture treatment
1846	Ether anesthesia demonstrated for surgery
1867	Lister publishes antiseptic surgical methods
1879	Macewen removes a meningioma in Glasgow
1884	Godlee removes a localized brain tumor in London
1887	Horsley removes a spinal cord tumor
Early 1900s	Cushing establishes modern neurosurgery
1927	Moniz introduces cerebral angiography
1937	Dandy clips a brain aneurysm
1947	Spiegel and Wycis perform stereotactic surgery in humans
Late 1960s	Leksell's Gamma Knife radiosurgery; microsurgery spreads
1971	First CT scan of a patient's brain
1974	Glasgow Coma Scale published
1990s	Detachable coils for aneurysms; endoscopic endonasal surgery
2010s onward	Deep learning applied to brain imaging and outcomes

How did microsurgery and minimally invasive techniques develop?

From the 1960s, the operating microscope transformed neurosurgery. M. Gazi Yaşargil, working in Zurich, developed many of the techniques and instruments of microneurosurgery, and in 1967 performed one of the first bypass operations connecting a scalp artery to a brain artery. Magnified, well-lit views let surgeons work between delicate structures instead of through them.

Other approaches avoid opening the skull at all. Lars Leksell developed stereotactic radiosurgery, delivering focused radiation to a precise target, with the first Gamma Knife in the late 1960s. In the early 1990s, detachable platinum coils made it possible to treat brain aneurysms from inside the blood vessels, a technique called endovascular coiling (see coiling vs clipping). In the 1990s surgeons also began removing pituitary and skull base tumors through the nose with endoscopes.

When was AI first used in healthcare?

The idea of artificial intelligence dates to the 1950s, and medicine was an early target. In the 1970s, Stanford's MYCIN system used hand-written rules to recommend antibiotics for infections. It performed well in tests but was never used in routine care. Through the following decades, medicine relied mostly on simpler statistical tools, such as risk scores built with regression.

Neurosurgery has a long tradition of these scores. The Glasgow Coma Scale (1974) and grading systems for subarachnoid hemorrhage turned clinical judgment into numbers that teams could share and compare.

The modern wave began in the 2010s, when deep learning made it possible for computers to learn patterns directly from images and large datasets.

How is AI used in neurosurgery today?

Machine learning now appears across the patient's path:

• Emergency imaging. In 2018 the FDA authorized software that flags suspected large vessel strokes on CT scans and alerts the stroke team, and similar tools now flag brain bleeds. The aim is to shorten the time to treatment.
• Diagnosis during surgery. Researchers have combined a laser imaging method called stimulated Raman histology with deep learning to identify brain tumor tissue in the operating room within minutes, rather than waiting for a pathology lab.
• Tracking the brain during surgery. The brain moves as fluid drains and tumor is removed, a problem called brain shift. Models that read intraoperative ultrasound can estimate that movement and help keep navigation accurate.
• Outcome prediction. Models estimate risks such as complications, tumor recurrence or the return of a treated aneurysm for an individual patient. Good models report how well they rank patients (the C-index) and whether their probabilities can be trusted (calibration).
• Explaining predictions. Methods such as SHAP show which factors drove a prediction, so clinicians can check the reasoning instead of trusting a single number.
• Brain-computer interfaces. Implanted electrodes paired with machine learning can translate brain signals into cursor movement or text for people with paralysis. These remain in clinical trials.

You can see several research examples of outcome prediction on our tools page.

Can AI perform brain surgery?

No. AI does not perform brain surgery. Robots are used in neurosurgery, mainly to guide the placement of spinal screws, biopsy needles and electrodes, but they follow a plan set by the surgeon and work under the surgeon's control. Autonomous surgical robots have been tested in research settings, including animal studies of soft tissue surgery, but nothing close to autonomous brain surgery is in clinical use.

What are the limits of AI in neurosurgery?

The same issues come up repeatedly:

• Small datasets. Many neurosurgical conditions are rare, so models are often trained on a few hundred patients from one hospital and may not work as well elsewhere.
• Validation. A model that looks accurate on the data it learned from can fail on new patients. Testing on separate data, ideally from other hospitals, is essential.
• Bias. Models learn from past care, including its gaps, and can perform differently across groups of patients.
• Prediction is not cause. A model can say a patient is at higher risk without showing that changing a treatment would lower it.
• Trust and regulation. Clinicians need to understand why a model made a prediction, and tools that inform care must meet regulatory standards.

These are the same lessons neurosurgery learned with every earlier technology: new tools help most when they are tested carefully and used by people who understand their limits.

Frequently asked questions

When was neurosurgery invented?

Skull surgery dates back more than 7,000 years, but neurosurgery as a modern specialty emerged in the late 1800s and early 1900s. Anesthesia, antiseptic technique and cerebral localization made it possible, and Harvey Cushing established it as a distinct field.

Who is called the father of modern neurosurgery?

Harvey Cushing is widely called the father of modern neurosurgery. He reduced the death rate of brain tumor surgery dramatically and trained many of the surgeons who built the specialty worldwide.

What was the first successful brain tumor surgery?

In 1879 William Macewen in Glasgow removed a meningioma that he had located from the patient's symptoms, and the patient survived. It is often cited as one of the first successful brain tumor operations.

Is AI used in brain surgery today?

Yes, but as a support tool. AI helps flag strokes and bleeds on scans, identify tumor tissue during surgery, track brain shift and estimate outcome risks. Surgeons make the decisions and perform the operations.

What is the future of AI in neurosurgery?

Expect more tools that work in real time during surgery, better outcome prediction from larger shared datasets, and continued progress in brain-computer interfaces. The main hurdles are validation across hospitals, explainability and regulation.