Samenvatting
Brain-computer interfaces (BCIs) are evolving toward higher electrode count and fully implantable solutions, which require extremely low power densities (<15mW cm−2). To achieve this target, and allow for a large and scalable number of channels, flexible electronics can be used as a multiplexing interface. This work introduces an active analog front-end fabricated with amorphous Indium-Gallium-Zinx-Oxide (a-IGZO) Thin-Film Transistors (TFTs) on foil capable of active matrix multiplexing. The circuit achieves only 70nV per sqrt(Hz) input referred noise, consuming 46µW, or 3.5mW cm−2. It demonstrates for the first time in literature a flexible front-end with a noise efficiency factor comparable with Silicon solutions (NEF = 9.8), which is more than 10X lower compared to previously reported flexible front-ends. These results have been achieved using a modified bootstrap-load amplifier. The front end is tested by playing through it recordings obtained from a conventional BCI system. A gesture classification based on the flexible front-end outputs achieves 94% accuracy. Using a flexible active front end can improve the state-of-the-art in high channel count BCI systems by lowering the multiplexer noise and enabling larger areas of the brain to be monitored while reducing power density. Therefore, this work enables a new generation of high channel-count active BCI electrode grids.
| Originele taal-2 | Engels |
|---|---|
| Artikelnummer | 2408576 |
| Aantal pagina's | 12 |
| Tijdschrift | Advanced Science |
| Volume | 12 |
| Nummer van het tijdschrift | 6 |
| Vroegere onlinedatum | 18 dec. 2024 |
| DOI's | |
| Status | Gepubliceerd - 10 feb. 2025 |
Bibliografische nota
Publisher Copyright:© 2024 The Author(s). Advanced Science published by Wiley-VCH GmbH.
Financiering
This publication is part of the project Smart\u2010Sense (with project number 17608) which is (partly) financed by the Dutch Research Council (NWO). This research was also supported by the Dutch Technology Foundation STW (PANDA project, grant 19072, MPB). Authors furthermore acknowledge the financial support of the EWUU Alliance (TU/e, WUR, UU, UMC Utrecht), project UEBIT.