The future
None of what follows is science fiction: all of it exists, published and at varying degrees of clinical maturity. Not all of it is part of my daily practice — yet. But following the field's evolution closely is part of practicing it seriously.
Brain–computer interfaces (BCI)
A brain–computer interface reads the electrical activity of neurons and translates it into action: moving a cursor, writing, speaking. In 2025 the field crossed an emotional threshold: implants that restored to paralyzed people the ability to communicate with a near-natural voice, decoding their intention to speak in near real time. Trials remain small and the challenges enormous — electrode durability, calibration, access — but the direction is unmistakable: the boundary between clinical neuroscience and assistive technology is dissolving. And for neurosurgery there is a matter of identity here: every one of these devices reaches the brain through a neurosurgeon's hands.
Adaptive deep brain stimulation
Deep brain stimulation — a fine electrode implanted in precise nuclei of the brain — has treated Parkinson's symptoms for decades. What is new is the adjective: 2025 saw the approval of the first adaptive system, which listens to the brain's own electrical activity and adjusts stimulation in real time, like a neurological thermostat. From a fixed stimulus programmed at the clinic to one that responds to the patient's physiological moment: it is the step from rigid prosthesis to dialogue. The lesson goes beyond Parkinson's — it opens a generation of devices that not only act on the brain, but listen to it.
Focused ultrasound (MRgFUS)
Hundreds of ultrasound beams cross the intact skull and converge, guided by real-time MRI, on a millimetric target. At that point — and only there — the energy produces a controlled therapeutic lesion. This is how essential tremor and certain Parkinson's symptoms are treated today: no incision, no implant, with the patient awake and responding during the procedure. Indications and targets have expanded over the past two years, and research is exploring its other face: transiently opening the blood–brain barrier to deliver drugs where they could not reach before. It is, strictly speaking, surgery without surgery.
Surgical robotics
In neurosurgery, the robot does not replace the surgeon: it lends qualities human physiology lacks. Absolute steadiness — no tremor, no fatigue —, sub-millimetric precision repeated a hundred times identically, and the ability to execute trajectories planned to the last detail. Today it assists mostly where geometric accuracy is everything: electrode placement for epilepsy or deep brain stimulation, spinal screws, stereotactic biopsies. The horizon — shared with microrobotics — is operating through ever smaller corridors with ever smarter instruments. The human hand does not withdraw: it moves up the hierarchy, from gesture to strategy.
Personalized medicine
Two brain tumors can look identical under the microscope and behave as different diseases. Molecular classification — mutations, methylation profiles, genetic signatures — revealed that difference and has already changed the rules: today a glioma's diagnosis is incomplete without its molecular biology, and that biology weighs on every subsequent decision — how much to resect, which adjuvant treatment, what prognosis to discuss honestly with the patient. Surgery itself takes on a new meaning: the specimen removed is no longer just tissue to excise; it is the key of information that decides everything that follows.
Digital twins
A digital twin is a computational replica of a real system — in this case, the patient. From a specific person's imaging, biomechanics and physiological data, a model is built on which one can rehearse: simulate an approach, anticipate how tissue will deform, compare strategies before committing to one. The simplest version already exists — 3D planning, training simulators, printed models of a patient's anatomy —; the ambition is larger: that every complex operation gets its dress rehearsal, and that error costs only computing time, never tissue.
New biomarkers
The dream is old: diagnosing and monitoring nervous-system disease without having to open it. Modern biomarkers are bringing it closer — circulating tumor DNA detectable in cerebrospinal fluid (the so-called liquid biopsy), blood proteins that betray neuronal injury, molecular signatures that anticipate treatment response. In neuro-oncology, liquid biopsy promises to answer questions that today demand surgery or uncertainty: is this growth on the MRI recurrence, or treatment effect? Still maturing, still being validated — but it points to a future where follow-up will be as molecular as it is radiological.
Surgical AI
Artificial intelligence entered neurosurgery through the door of imaging — detecting, segmenting, classifying — and is advancing into more delicate territory: predicting an individual patient's surgical risk, anticipating complications, assisting trajectory planning, even analyzing surgical video to recognize structures and procedural phases in real time. Its promise is not to replace clinical judgment but to feed it better inputs; its risk, being adopted without the validation demanded of any other medical tool. The operative question of the decade will not be "AI, yes or no?" but "validated how, supervised by whom?".