The long-term goal of brain-computer interface (BCI) research is to establish a new mode of communication for individuals who have lost some or all voluntary muscle control due to injury or degenerative diseases such as amyotrophic lateral sclerosis, multiple sclerosis, and severe cerebral palsy. If all voluntary muscle control is lost, a locked-in syndrome results in which a person is unable to communicate with the outside world. BCIs could potentially provide a way for these individuals to communicate with their caregivers and to control devices such as televisions, wheelchairs, and robot assistants. While BCI technology holds great promise, non-invasive BCI systems are not yet practical, primarily due to limitations in signal quality provided by current electroencephalogram (EEG) scalp electrodes. This project will explore initial steps towards a research plan that will transform BCI technology in ways that will enable breakthroughs in the reliability and accuracy of BCI applications. After years of limited advances in BCI accuracy and reliability, project outcomes will accelerate the design of new BCI applications to significantly improve the quality of life for many persons who are in dire need of help. The project will also play a strong role in the interdisciplinary education of computer science and biomedical students.
In this exploratory project, equipment will be acquired to enable the recording of high-quality EEG signals generated by a new non-invasive tripolar concentric ring electrode EEG sensor being developed by Dr. Walter Besio of the University of Rhode Island, which enables the recording of brain activity with much more spatial and temporal precision than what is possible with conventional EEG electrodes. The EEG data thus obtained will provide the information needed by novel deep learning algorithms to translate brain activity to intended arm and hand movements, and experiments will be performed to demonstrate the feasibility of detecting real and imagined individual finger movements. The belief for decades has been that detecting finger movements requires invasive, implanted electrodes to avoid degradation of brain signals as they pass through cerebral-spinal fluid, skull and skin. This research will be the first to try a new end-to-end deep learning approach to translating brain activity to arm and hand movements.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
