Sponsored
🚀 Future

Injectable Mesh Electronics Could Enable Long-Term Neural Recording Without Surgery, Researchers Say

A blue capsule connected to thin lines passing through a series of glass panels

In the spring of 2015, in a chemistry lab on Oxford Street in Cambridge, Massachusetts, a postdoctoral researcher named Jia Liu loaded a glass syringe with a liquid that did not look like it belonged anywhere near a brain. Suspended in saline was a mesh: a lattice of flexible polymer struts and gold interconnects, folded and crumpled so it could pass through a needle narrower than the ones used for a flu shot. Liu was working in the lab of Charles Lieber, a Harvard chemist who had spent years building nanoscale electronic structures, and the two of them were testing something that sounded almost absurd on paper. Could you deliver a functioning electronic circuit into living tissue the same way you’d deliver a vaccine?

They could. Injected into saline-filled cavities and later into the brains of live mice, the mesh unfurled on its own once it left the needle, relaxing into a loose three-dimensional scaffold that threaded itself between cells rather than pushing them aside. The result, published in Nature Nanotechnology in 2015, was less a new electrode than a new category of object: electronics soft enough, porous enough, and structurally similar enough to brain tissue that the tissue did not seem to notice it was there.

A Syringe Instead of a Craniotomy

Every brain implant in clinical use today, from the Utah Array used in BrainGate trials since 2004 to the electrode threads Neuralink began implanting in human patients in January 2024, gets into the brain the same way: through a hole cut in the skull. The craniotomy is not incidental to the risk profile of these devices. It is the risk profile. Infection, scarring at the surgical site, and the trauma of inserting a rigid probe into soft tissue all trace back to that same opening step, long before anyone asks how well the electrode actually records.

Mesh electronics sidesteps that step entirely. Because the scaffold is delivered in a collapsed, liquid-suspended form, it can in principle pass through a needle track rather than a surgical window, the same way a contrast dye or a chemotherapy infusion reaches its target. Liu, who has since built his own lab at the Harvard John A. Paulson School of Engineering and Applied Sciences, has continued pushing the approach, developing variants capable of recording from the brain, peripheral nerves, and even lab-grown neural organoids using the same injection logic.

The needle isn’t delivering a probe. It’s delivering a network that happens to arrive folded.

What makes this more than a delivery trick is what happens after the mesh arrives. A conventional electrode is a foreign object the moment it lands, a rigid shard sitting inside tissue that is, by comparison, almost liquid. Mesh electronics is built to fail that comparison on purpose. Its struts are thin enough, and its open lattice sparse enough, that neurons and glial cells grow through the mesh rather than around it, effectively absorbing the electronics into the tissue architecture instead of walling it off.

Why Stiffness, Not Resolution, Has Been the Real Limit

Electrode engineers have spent decades chasing more channels, finer tips, denser arrays. Lieber’s group argued that channel count was never the binding constraint. The binding constraint was mechanics. Silicon and metal electrodes have an elastic modulus measured in gigapascals. Brain tissue sits in the range of a few kilopascals, roughly six orders of magnitude softer. Every heartbeat, every breath, every small shift of the skull produces micromotion between a stiff implant and the soft tissue around it, and that friction is what triggers the brain’s standard response to a splinter: astrocytes and microglia converge on the site, wall it off in scar tissue, and the signal that was crisp on day one degrades into noise by month three.

A 2017 paper from Lieber’s group in Nature Methods reported something that conventional probes had never managed: stable recordings from the same individual neurons in mice for as long as eight months, with minimal evidence of the glial scarring that chokes off signal from rigid implants. That duration matters more than any spike in channel count. A brain-computer interface that works beautifully for six weeks and then degrades is a laboratory curiosity. One that holds a stable signal for the better part of a year starts to look like a tool you could build a therapy around.

A rigid electrode treats the brain like a circuit board. A mesh treats it like tissue.

The mechanism is almost disappointingly simple once stated plainly. Match the stiffness of the implant to the stiffness of the organ, and the organ stops fighting it. The idea has precedent elsewhere in bioelectronic medicine, where engineers building vagus nerve cuffs and peripheral nerve interfaces have made the same bet on soft, compliant materials over rigid ones. Mesh electronics pushes that logic further than almost anything else in the field, because it combines mechanical softness with a delivery method that requires no incision at all.

How the Approaches Stack Up

Mesh electronics is not competing in a vacuum. It sits alongside a handful of other strategies for getting electrodes into or near the brain with less trauma than a traditional craniotomy, each trading invasiveness against signal quality and regulatory maturity in different ways.

Technology Delivery Method Invasiveness Current Status
Utah Array (Blackrock Neurotech) Direct implantation via craniotomy High, open skull surgery Investigational, used in BrainGate human trials since 2004
Neuralink Robotic thread insertion via craniotomy High, open skull surgery FDA IDE approved 2023, first human implant January 2024
Synchron Stentrode Endovascular catheter via jugular vein Moderate, no craniotomy, still a catheter procedure FDA IDE approved 2021, US trial ongoing
Mesh electronics (Lieber, Liu) Syringe injection Low, needle puncture only Preclinical, demonstrated in rodent and organoid models

The table understates how far mesh electronics still has to travel. A catheter-delivered Stentrode has already reached human patients. Mesh electronics has not, and the jump from a mouse cortex to a human skull involves questions of scale, targeting, and signal routing that have not been solved in print. But the trend line is unambiguous. Each successive approach asks the body to tolerate less. The craniotomy gives way to the catheter, and the catheter, if mesh electronics works out, gives way to the needle.

A Research Program Interrupted

The story of mesh electronics cannot be told without the strange detour its principal inventor took through federal court. In January 2020, Charles Lieber was arrested and charged with lying to federal investigators about his ties to Wuhan University of Technology and his participation in China’s Thousand Talents Plan, a recruitment program, while receiving tens of millions of dollars in funding from the National Institutes of Health and the Department of Defense. He was convicted in December 2021 on two counts of making false statements. The case, playing out while Lieber was also being treated for lymphoma, effectively ended his ability to run a federally funded laboratory at Harvard.

The technology survived him. Jia Liu, who had co-authored the foundational mesh electronics papers as a postdoc, had already established an independent lab by the time the indictment came down, and he has since extended the work into new territory: injectable electronics for peripheral nerves, implantable electronics for lab-grown neural organoids, and continued refinement of long-term chronic recording in rodents. The Lieber case is a reminder that the institutional path of a technology and the scientific path of a technology are not the same thing. One can falter while the other keeps moving.

What Happens When the Barrier Drops From an Operating Room to an Office Visit

Every real advance in neural interface technology has followed the same arc: what once required a neurosurgeon eventually requires a technician, and what once required a hospital eventually requires an outpatient visit. Deep brain stimulation moved from experimental neurosurgery to a standard Parkinson’s treatment performed in dedicated surgical centers. Vagus nerve stimulators shrank from implanted hardware to, in some applications, handheld transcutaneous devices a patient can use at a kitchen table. Mesh electronics represents the next step down that same curve, and it is worth being precise about what changes when it arrives.

A craniotomy is, among other things, a natural ethical and practical filter. It requires informed consent documents, a surgical team, a hospital stay, and a risk calculus serious enough that almost nobody undergoes one casually. A syringe does not carry the same weight. If injectable mesh electronics eventually reaches people, the question of who gets a long-term neural recording interface stops being answered by “who has a condition severe enough to justify brain surgery” and starts being answered by something closer to “who has access to the injection and can afford it.” That is a much larger population, and a much less self-selecting one.

Twenty years from now, the hard problem may not be the needle. It may be deciding who gets to use it.

There is also a permanence question nobody has fully answered. A Stentrode sits inside a blood vessel and can, in principle, be retrieved by catheter. A craniotomy-delivered array sits in a defined location and can be surgically removed. Mesh electronics is designed to do the opposite of sit still: its entire value proposition is that neurons and glia grow through it and incorporate it into the surrounding tissue. That is exactly what gives it its biological stability, and exactly what makes it unclear how, or whether, it comes back out. A technology engineered to disappear into the brain raises a different kind of informed consent problem than one engineered to be extracted on command.

None of this is an argument against the research. Bioelectronic medicine has spent twenty years learning that the body’s own electrical signaling can be read and redirected with enough precision to treat epilepsy, depression, and inflammatory disease without a single new molecule. Injectable mesh electronics extends that same project toward something that has always been medicine’s hardest constraint: duration. A signal worth reading for eight months is a signal worth building a diagnosis, or a therapy, or eventually a continuous relationship with a patient’s own nervous system, around. The philosophical weight of that is not abstract. It is the difference between a brain implant as an occasional clinical event and a brain implant as a standing condition of a person’s life, something closer to a pacemaker than a one-time procedure.

The field has not yet had to answer what a neural interface owes the person wearing it once it can no longer be removed, or how continuous outpatient neural data should be stored, who can request it, and for how long it remains meaningful after the person it came from has changed. Those questions used to be deferred, reasonably, on the grounds that so few people would ever carry an implant that the edge cases could wait. A syringe changes that math before a single human patient has received one.

Mesh electronics is still a rodent technology, demonstrated in mice and in organoids, with no published human trial and no clear regulatory pathway yet defined for a device that is, strictly speaking, not implanted at all in the surgical sense. The distance between a mouse cortex and a human skull is real, and the people who built this technology have been careful not to promise more than the data support. But the direction of travel across twenty years of neural interface research, from craniotomy to catheter to needle, points at a specific and answerable question: once getting a long-term recording device into a human brain requires nothing more than an injection, what, exactly, is going to be the thing that still says no.

Credit: Google DeepMind on Unsplash

bioelectronic medicinebrain-computer interfaceCharles Lieberchronic neural recordinginjectable mesh electronicsmesh electronics brain implantminimally invasive brain-computer interfaceneural recordingsyringe-injectable neural electrodes
Facebook
Twitter
LinkedIn
Stay charged
The electric pulse of discovery, in your inbox.

One weekly email. The most fascinating stories at the intersection of biology, electricity, and the future. No noise.