Brain-computer interfaces (BCIs) hold the potential to regain lost abilities, such as the ability to speak for paralyzed patients or see for the blind. A big problem, though, is that the hardware is intrusive; most high-performance BCIs use arrays of electrodes that go into the brain, which is a very invasive operation that can cause long-term tissue damage if not done properly. A recent research published in Nature Biomedical Engineering details a thin-film microelectrode array that can detect and activate neural activity through over a thousand channels—all without invasive craniotomy or cortical punctures.
Precision Neuroscience Corporation’s research team has created a cortical surface array of 1,024 electrodes made of brain-conforming, flexible thin-film material. To avoid removing substantial amounts of bone, a “cranial micro-slit” technique can be used to insert the electrodes beneath the skull through a small opening that is less than one millimeter wide. The system has proven to be capable of recording neural activity at high resolution and stimulating specific areas of the brain in preclinical studies using human cadavers and Göttingen minipigs. Now, it has been tested on five patients undergoing neurosurgery. A safer, reversible alternative to traditional penetrating arrays, the implants can be removed without leaving visible cortical damage, which is crucial.
Minimally invasive delivery, increased electrode density
When it comes to epilepsy mapping, the standard clinical electrocorticography (ECoG) grids usually don’t have more than 100 electrodes spaced a few millimeters apart. The new thin-film devices have a tremendous improvement in coverage and density, allowing thousands of electrodes to be deployed simultaneously with sub-millimeter resolution across large areas of cortex. This level of detail allows researchers to track the brain’s activity down to the cellular level, capturing not just general patterns but also the precise spatial markers of sensation, movement, and speech.
Despite the 300-micron spacing, the researchers asserted that they detected distinct signals being picked up by nearby electrodes, particularly at higher frequencies. This indicates that the cortical surface is home to a wealth of information contained in the brain’s electrical activity that was previously uncaptured by arrays.
The advancement is supported by a surgical innovation known as the cranial micro-slit. In order to allow the wafer-thin electrode film to slide into the subdural space, surgeons create a slit that is just wide enough rather than drilling a large craniotomy. In less than 20 minutes, the array is placed over functional areas like the motor, sensory, or visual cortex under the guidance of imaging and endoscopy. The process worked quickly, accurately, and consistently on cadaver heads and animals. Even after 42 days of implantation, histological examination revealed no signs of tissue damage, and the devices could be carefully removed without causing any damage to the cortical surface.
Additionally, arrays covering multiple cortical regions or even both hemispheres can be linked together thanks to the modular design. Scientists implanted over 2,000 electrodes into pigs all at once to record their motor and sensory cortices as they walked on a treadmill.
Recording and stimulation in one
With such a large array of electrodes, machine learning algorithms can decipher neural activity with remarkable precision. The system was able to correctly categorize pigs’ neural activation patterns (shown by stimulation at dozens of points on the face) more than 90% of the time in somatosensory tests. Arrays placed over the occipital cortex reliably detected visual evoked potentials, and signals recorded during locomotion allowed researchers to distinguish between the animal’s head, limbs, and resting states.
A preliminary human trial tested the technology in neurosurgeries, where patients having tumors removed or epilepsy surgery had the arrays transiently implanted on their brain surface. In awake patients, the device captured high-resolution speech-related activity from language areas, enabling researchers to detect the onset of spoken words with nearly 80% accuracy using just a few minutes of training data. The high-density arrays showed more precise maps of functional borders than the standard 4-electrode strips.
The arrays can do more than just “listen”; they can also “talk” back to the brain. Electrodes can modulate cortical activity by delivering tiny pulses of current, acting as stimulators. In pigs, sub-millimeter control was demonstrated by focal stimulation of the visual cortex, which caused changes in high-frequency activity that were localized to just a few millimeters around the site. For future closed-loop BCIs to decode intent and provide sensory feedback, such as touch in a prosthetic limb or visual percepts in a cortical vision prosthesis, this bidirectional capability is crucial.
A path toward clinical translation
The promise of BCIs has always been balanced against their risks. Penetrating electrodes offer the richest signals but are prone to scarring, instability, and difficult removal. Surface arrays like ECoG are safer but historically too coarse for high-performance decoding. By bridging that gap, thin-film micro-ECoG arrays could dramatically expand the clinical viability of BCIs.
Equally important is scalability. Implanting thousands of penetrating electrodes would normally mean more tissue disruption and longer surgery times. Here, electrode count scales without increasing invasiveness: more arrays can be inserted through the same slit, covering larger portions of cortex in minutes. That makes it conceivable to blanket wide swaths of the human brain with high-density sensors, unlocking rich neural information while maintaining safety.
The current study demonstrates feasibility in acute and short-term settings. Long-term human trials, with fully integrated wireless systems, will be needed to prove durability and clinical benefit. Still, the pilot data are compelling: intraoperative use showed that real-time decoding of speech and motor activity is possible, and stimulation can modulate cortical signals in predictable ways.
The researchers argue that safety, scalability, and reversibility are essential for BCI technologies to reach the millions of patients who could benefit, from those with paralysis to those requiring high-precision neurosurgery. The thin-film array represents a major stride toward that vision, offering a practical and less invasive path to high-bandwidth communication between brain and machine.
For now, the technology remains experimental. But if future studies confirm its safety and utility in chronic implants, thin-film cortical arrays may become the foundation for next-generation BCIs—ones that can be implanted in a routine neurosurgical procedure and one day help restore lost human abilities without requiring the brain to be pierced by metal.
