The dominant path to a high-resolution brain-computer interface runs through a craniotomy: open the skull, place electrodes on or in the cortex, and close. It works, but the surgery itself is the barrier — invasive, high-risk, and hard to justify for anything short of severe indications. A patent issued to William Marsh Rice University on July 7, 2026 describes a different route to the same neural real estate. Instead of going through the skull, it goes through the fluid that already surrounds the brain and spinal cord: the subarachnoid space.

The patent, US12673202B2, “Cortical subarachnoid and intra ventricular brain interfaces,” is directed to neural interface devices and methods that access the subarachnoid space to enable minimally invasive modulation and recording of neural structures. In the disclosed method, access begins with a lumbar puncture — the same spinal-tap procedure used to sample cerebrospinal fluid — after which a microcatheter is advanced through the spinal subarachnoid space. Because that space is a continuous fluid channel wrapping the central nervous system, a sufficiently thin, steerable catheter can travel it toward intracranial targets without cutting through brain tissue or bone.

The present disclosure is directed to neural interface devices and methods that accesses the subarachnoid space to enable minimally invasive modulation and recording of neural structures. Exemplary embodiments may comprise an implantable pulse generator and a microelectrode catheter. In particular embodiments, the microelectrode catheter comprises one or more stimulating and recording electrodes. Exemplary embodiments may also include methods comprising performing a lumbar puncture to access the spinal subarachnoid space and advancing microcatheter through the spinal subarachnoid space.— Cortical subarachnoid and intra ventricular brain interfaces, US12673202B2

The catheter and its electrodes

Claim 1 defines the device in two parts: an implantable pulse generator and a microelectrode catheter carrying stimulating and recording electrodes, with those electrodes configured for implantation into the spinal and intracranial subarachnoid space or into the ventricles of the brain. The single instrument therefore does two jobs — it can record neural activity and it can deliver stimulation — and it can be positioned either against the cortex from within the surrounding fluid or inside the ventricular system deeper in the brain. The disclosed stimulation parameters are specific: amplitudes of at least 12.0 volts delivered in pulses on the order of 250 microseconds.

Reaching the cortex from within the cerebrospinal fluid rather than through it is what makes the approach minimally invasive. The subarachnoid space is the thin, fluid-filled layer between the arachnoid and pia mater that follows the surface folds of the brain and runs the length of the spinal cord as one continuous compartment. A device navigated along that compartment can be brought adjacent to cortical structures — or, per the claim, into the ventricles — without the retraction and resection that open surgery entails. The trade is one of access versus proximity: the electrodes sit in fluid near their targets rather than being driven into tissue, and the disclosed stimulation amplitudes reflect the need to drive a signal across that intervening space.

The harder problem for any long-term implant reached this way is power. Running a wire out of the subarachnoid space to a battery would undercut the minimally invasive premise. The patent addresses this with magnetoelectric wireless powering. In one claim, an external field transmitter emits an alternating magnetic field in the 20 kHz to 1 MHz range, and a magnetoelectric film on the implant is driven at its mechanical resonance — the magnetic field makes the film vibrate, and that vibration produces a voltage the implant uses for power. The same physics carries data back out: the disclosure describes a magnetoelectric backscatter communication protocol, so the implant can report without its own radio. That magnetoelectric-bioelectronics approach is a signature line of work for lead inventor Jacob Robinson, listed alongside co-inventors Peter Kan, Joshua Chen, and Abdeali Dhuliyawalla. Magnetoelectric films are attractive for implants because low-frequency magnetic fields pass through tissue with little absorption and without the heating that limits radio-frequency and ultrasound alternatives, and because driving the film at its mechanical resonance maximizes the voltage recovered from a modest external field. Reusing that same coupling for backscatter communication means the implant needs neither a battery nor a conventional transmitter — two of the bulkiest components in a conventional neurostimulator.

Academic IP, not a product roadmap

This is a university patent, and it reads as translational research rather than a commercial launch. It sits within a small same-week cohort of Rice bioengineering grants that together sketch the breadth of the school's estate rather than a single product line. US12673132B2, out of the Mikos lab, is directed to extrusion printing of biocompatible scaffolds — 3D-printed porous constructs for bone and cartilage. US12673072B2, from the Veiseh lab, is directed to encapsulated cells expressing IL-12, an implantable cell-encapsulation construct for delivering that cytokine. Neural interfaces, printed tissue scaffolds, and encapsulated cell therapy are different disciplines, but each is an implantable-bioengineering bet.

What ties the subarachnoid interface to the field's state of the art is its combination of two independently active research fronts: endovascular- and CSF-route neural access that avoids the craniotomy, and battery-free implants powered by fields that penetrate tissue safely. The device described in US12673202B2 puts both in one system — a catheter-delivered electrode array reaching cortical and ventricular targets through the body's own fluid corridors, running on power beamed in as a magnetic field. As an issued grant from a university, it stakes out that combination in the patent record; whether and how it reaches patients is a separate, longer question the document does not answer.