Stentrodevs.Connexus Brain-Computer Interface
Synchron Stentrode vs Paradromics Connexus: Endovascular vs Intracortical
Both have FDA IDEs. One uses a catheter; the other uses a craniectomy. They're not competing for the same patient.
Synchron's Stentrode is a 16-channel endovascular ECoG array delivered via the jugular vein in an outpatient catheter procedure — the minimally-invasive bet, with a peer-reviewed safety record from SWITCH (Australia) and an active US COMMAND trial. Paradromics' Connexus is a 421-electrode intracortical implant that scales modularly to 1,600+ channels, designed around speech restoration and cleared for first-in-human under the Connect-One IDE in November 2025. Same regulatory status; near-opposite hardware bets.
Published · Updated
Side-by-side specs
| Specification | Stentrode | Connexus Brain-Computer Interface |
|---|---|---|
| Classification | ||
| Invasiveness | minimally-invasive | invasive |
| Primary modality | LFP | single-unit |
| Direction | read | read |
| Electrodes | ||
| Total channels | 16 | 421 |
| Recording channels | 16 | 421 |
| Electrode type | stentrode-mesh | penetrating-shank |
| Prep time | — | — |
| Acquisition | ||
| Sampling rate | — | — |
| ADC resolution | — | — |
| Connectivity | ||
| Protocols | transcranial-inductive, proprietary-rf | transcranial-inductive |
| Power | ||
| Battery life (active) | — | — |
| Physical | ||
| Weight | — | — |
| Software | ||
| Raw data access | No | No |
| LSL support | — | — |
| SDK | ||
| Has SDK | No | No |
| Open source | No | No |
| Regulatory | ||
| FDA status | investigational-device-exemption | investigational-device-exemption |
| CE mark | — | — |
| Pricing | ||
| MSRP | — | — |
| Subscription required | — | — |
| Warranty | — | — |
Verdict by axis
channel-count
Connexus Brain-Computer Interface
421 vs 16 — and Connexus is designed to scale modularly to 1,600+ channels across 4 implants.
Confidence: high
signal-quality
Connexus Brain-Computer Interface
Penetrating intracortical electrodes resolve individual neuron action potentials. Vascular ECoG sees aggregated cortical activity through the vessel wall. Connexus's pre-clinical 200+ bits/s ITR claim — if it holds in humans — would be substantially above what 16-channel vascular ECoG can deliver.
Confidence: medium
invasiveness
Stentrode
Stentrode requires no craniectomy — it's delivered via the jugular vein in an outpatient procedure. Connexus requires open surgery and multi-module placement on cortex.
Confidence: high
safety
Stentrode
Stentrode has a published peer-reviewed safety record from SWITCH plus an active COMMAND cohort. Connexus has not yet implanted humans (Connect-One IDE granted 2025-11-20).
Confidence: medium
reversibility
Stentrode
Endovascular devices can in principle be retrieved via the same vascular route. Multi-module penetrating cortical implants cannot be cleanly explanted without further injury.
Confidence: medium
research-credibility
Stentrode
Stentrode has peer-reviewed clinical results in JAMA Neurology (2023). Connexus's published evidence is currently pre-clinical / manufacturer-cited.
Confidence: high
clinical-evidence
Stentrode
Synchron has a multi-patient clinical safety dataset from SWITCH and active follow-up from COMMAND. Paradromics is at the threshold of first-in-human as of 2026-05.
Confidence: high
capabilities-cursor
Stentrode
Stentrode has demonstrated cursor and click control in patients living at home. Connexus has not yet been implanted in humans.
Confidence: high
capabilities-typing
Connexus Brain-Computer Interface
Connexus's channel count is in the range that has produced high-WPM decoding on other intracortical hardware. Stentrode's 16-channel ceiling caps the achievable rate. (Both are projections — neither has matched the published Utah-array results yet.)
Confidence: low
capabilities-prosthetic
Connexus Brain-Computer Interface
Multi-DOF prosthetic control historically requires hundreds of intracortical channels, putting Connexus's architecture in range and Stentrode's out of range — projection only.
Confidence: low
availability
tie
Neither is commercially available. Synchron's COMMAND is actively recruiting; Paradromics's Connect-One is just starting.
Confidence: high
stimulation
tie
Both are read-only as currently studied.
Confidence: high
support
Stentrode
Stentrode is implanted by interventional neuroradiologists who already perform similar stent procedures daily. Connexus requires neurosurgical teams trained on a new multi-module workflow.
Confidence: medium
Pros & cons
Stentrode
In favor
- Minimally invasive — no craniectomy, ~2-hour catheter procedure
- Implanted by interventional neuroradiologists already trained in stent procedures
- Strong peer-reviewed safety record (SWITCH + COMMAND)
- First permanently-implanted BCI under a US FDA IDE
- Procedural model fits existing hospital workflows
- Likely faster path to broad clinical deployment if endpoints continue to hold
Against
- Only 16 channels — orders of magnitude below intracortical devices
- Vascular ECoG is lower-resolution than penetrating recording
- Limited to motor-cortex regions accessible via the sagittal sinus
- Unlikely to reach the channel counts speech decoding requires
Connexus Brain-Computer Interface
In favor
- 421 channels per module — 26× Stentrode's count
- Modular: scales to 1,600+ channels across 4 implants
- Custom on-implant ASIC with signal conditioning + telemetry
- FDA Breakthrough Device + IDE; targets the high-value speech-restoration application directly
- Wireless transcutaneous power — no skin-penetrating connector
- Pre-clinical 200+ bits/s ITR claim is industry-leading if reproduced in humans
Against
- Pre-clinical / first-in-human stage — limited published human data so far
- Most internal hardware specs proprietary / unverified
- Multi-module implantation increases surgical exposure
- Requires open craniectomy and a neurosurgical workflow that doesn't exist at scale
Recommendations by use case
| Use case | Pick | Why |
|---|---|---|
| Cursor + click for severe paralysis (today) | Stentrode | Stentrode has published clinical data covering exactly this paradigm in patients living at home. |
| Speech restoration | Connexus Brain-Computer Interface | Connexus is purpose-built for this application; vascular ECoG at 16 channels is unlikely to deliver high-quality continuous speech. |
| Digital switch / AAC for ALS | Stentrode | Demonstrated in SWITCH and COMMAND with a peer-reviewed safety record. |
| Multi-DOF prosthetic / robotic-arm control | Connexus Brain-Computer Interface | Historically requires hundreds of intracortical channels — within Connexus's design envelope, outside Stentrode's. |
| Patients ineligible for craniotomy | Stentrode | Endovascular delivery sidesteps surgical risks craniectomy patients may not tolerate. |
| Maximum surgical reversibility | Stentrode | Endovascular devices can in principle be retrieved via the same route. |
| Maximum future bandwidth ceiling | Connexus Brain-Computer Interface | 1,600+ scalable channels is multiple orders of magnitude beyond Stentrode's design. |
| Hospital-scale deployability | Stentrode | Neurointerventionalists exist at thousands of hospitals; multi-module Connexus surgery is a new workflow. |
| Long-term recording stability research | Stentrode | Vascular contact may avoid the gliosis that degrades penetrating arrays — though long-term Connexus data is needed. |
| Buying one for personal use | Neither | Both are investigational devices in clinical trials. Neither is purchasable. |
Frequently asked
›Are these devices competing for the same patients?
Partly. Both target severe motor impairment from ALS, SCI, stroke, and locked-in syndrome. But the use cases diverge: Stentrode's published wins are in cursor / click / digital-switch control, while Connexus is purpose-built for speech restoration — an application Stentrode's 16 channels almost certainly cannot deliver.
›Why does Connexus need so many more channels?
Speech decoding is bandwidth-hungry. The 2023 Stanford speech BCI used 256 intracortical channels and reached ~62 WPM; Paradromics is publicly targeting higher rates by scaling channel count further. Vascular ECoG can detect attempted-movement signals, but does not resolve the population activity that current speech decoders rely on.
›Has Connexus been implanted in humans?
Not yet. Paradromics received an FDA IDE for the Connect-One first-in-human study on 2025-11-20. As of 2026-05, no human implants have been publicly announced.
›Is Stentrode safer because it's less invasive?
Likely, on the procedural-risk axis — endovascular delivery avoids craniectomy. But 'less invasive' isn't synonymous with 'safer over years' in BCIs: vessel-wall integration carries its own risk profile (thrombosis, vessel patency), and long-term comparative data does not yet exist.
›Which one will reach commercial approval first?
Synchron's path is generally seen as faster — endovascular procedures fit existing neurointerventional workflows, and SWITCH / COMMAND already have multi-year clinical data. Connexus is earlier-stage but is targeting a higher-value indication with FDA Breakthrough designation. Both are pre-PMA as of 2026-05.
Bottom line
These devices solve overlapping problems with opposite tradeoffs. Stentrode is the safer, more clinically-validated option today, and the right answer for cursor-class assistive applications. Connexus is the higher-bandwidth bet on speech restoration, but the human evidence has not arrived yet. Both will likely succeed in different patient populations.