Researchers from the Georgia Institute of Technology have created a new electroencephalography (EEG) system from new classes of nanomembrane electrodes with flexible electronics and a deep learning algorithm. The system could help disabled people wirelessly control electric wheelchairs, computers, small robotic vehicles and more without an EEG electrode cap.
The new system is a fully portable brain-machine interface (BMI). BMIs allow patients with ALS, chronic stroke or other motor disabilities to control robotic prosthetics. The system could also help non-invasively identify biomarkers of Alzheimer’s.
To gather brain signals, doctors typically use an electrode-studded hair cap outfitted with wet electrodes and adhesives that is wired to a computer. The majority of the participant’s scalp must be covered for the system to receive signals. This can be cumbersome and annoying, so many patients choose not to use these systems daily.
The new system is composed of three parts: flexible, hair-mounted electrodes; ultra-thin, nanomembrane electrodes; and soft, flexible circuitry equipped with Bluetooth. To use the system, a skin-like electrode is placed below the ear without adhesives or gel.
The system reportedly reduces noise and interference and has higher transition rates than current technologies. EEG data from the brain is processed in the flexible circuitry, which is delivered to a tablet through Bluetooth up to 15 m away. Since EEG signals vary between individuals, a deep-learning algorithm analyzes and decodes the signals to turn them into actions.
Researchers tested their system on six able-bodied human subjects, but plan to test the system with disabled subjects in the future. The participants successfully used the new system to control an electric wheelchair and a small robotic vehicle without input, like a joystick or keyboard.
Next, the team wants to improve the electrodes and enable the system to control more electronics. There are plans to innovate fully elastomeric, wireless, self-adhesive electrodes that are mounted on a scalp without any headgear, and miniaturize the electronics so they can be incorporated into other studies.
The study was published in Nature Machine Intelligence.