Biopotential signals contain essential information for assessing functionality of organs and diagnosing diseases. We present a flexible sensor, capable of measuring biopotentials, in real time, in wireless and fully-passive manner. The flexible sensor collects and transmits biopotentials to an external reader without wire, battery, or harvesting/regulating element. The sensor is fabricated on a 90 μm-thick polyimide substrate with footprint of 18 × 15 × 0.5 mm3. The wireless fully-passive acquisition of biopotentials is enabled by the RF (Radio Frequency) microwave backscattering effect where the biopotentials are modulated by an array of varactors with incoming RF carrier that is backscattered to the external reader. The flexile sensor is verified and validated by emulated signal and Electrocardiogram (ECG), Electromyogram (EMG), and Electrooculogram (EOG), respectively. A deep learning algorithm analyzes the signal quality of wirelessly acquired data, along with the data from commercially-available wired sensor counterparts. Wired and wireless data shows <3% discrepancy in deep learning testing accuracy for ECG and EMG up to the wireless distance of 240 mm. Wireless acquisition of EOG further demonstrates accurate tracking of horizontal eye movement with deep learning training and testing accuracy reaching up to 93.6% and 92.2%, respectively, indicating successful detection of biopotentials signal as low as 250 μVPP. These findings support that the real-time wireless fully-passive acquisition of on-body biopotentials is indeed feasible and may find various uses for future clinical research.
|Original language||English (US)|
|Journal||Biosensors and Bioelectronics|
|State||Published - Aug 15 2019|
- Deep learning
ASJC Scopus subject areas
- Biomedical Engineering