Sewing-machine polymer electrode threads (UCSF and Berkeley)
Sewing-machine polymer electrode threads
A thin-film electrode system paired with a robotic inserter. A stiff needle engages a loop on a flexible thread, inserts it at an individually chosen location, then withdraws. The thread stays in the tissue. The 2019 bioRxiv preprint describes the electrode array, insertion needle and robot as separate parts of the system.
Hardware
| Field | Preprint detail |
|---|---|
| Fabricated array | 64 individual threads |
| Sites per thread | One recording or stimulating site |
| Thread substrate / conductor | Polyimide / platinum |
| Shank width | 16 µm |
| Conductor trace width | 4 µm |
| Thread length | 27.25 mm in Figure 2 |
| Bondpad region | 4.2 × 7.7 mm in Figure 2 |
| Insertion needle | 25 µm, discussed in comparison with injectable mesh |
| Reported insertion cycle | Less than nine seconds per thread |
Some threads include barbs. The loop, taper, barbs, site and long lead mean this is not adequately represented by a straight rectangular ribbon. No full 3D model is added here.
What was demonstrated
The team recorded extracellular activity in four rats during development. One rat had 24 implanted electrodes recorded over two months. Single-unit activity was reported on 39% of implanted electrodes overall: 3 of 22, 2 of 13, 7 of 12 and 16 of 24 across the four animals.
The 64-thread fabricated design is not a claim that all 64 sites were implanted or recorded in every animal.
What did not work cleanly
Three recording animals lost their implants prematurely. The authors describe surgical reliability limits including blood obscuring targets, needle-cannula clogging, microdurotomy depth and implant durability. Histology showed neuronal loss in lesion cores and increased GFAP around insertion sites. The paper reports damage as well as recordings; the device is not catalogued as damage-free.
The preprint states that its probes had not shown the longevity and stability reported for some other flexible probes. That is a limit of this report, not proof that the project was abandoned. No later clinical outcome is asserted here.
Provenance
The preprint lists UCSF Physiology, the joint Berkeley/UCSF Bioengineering graduate group, Berkeley Electrical and Computer Engineering and the Chan Zuckerberg Biohub among its affiliations. It was posted March 14, 2019 and is explicitly marked not certified by peer review. This entry is not the Neuralink N1 device or a claim about its origins.
Source
- Hanson TL, Diaz-Botia CA, Kharazia V, Maharbiz MM, Sabes PN. The “sewing machine” for minimally invasive neural recording. bioRxiv 578542 (2019). DOI: 10.1101/578542. Full primary preprint, methods and results.