Meeting brain-computer interface user performance expectations using a deep neural network decoding framework

• Published in: Nature medicine Vol. 24; no. 11; pp. 1669 - 1676
• Main Authors: Schwemmer, Michael A, Skomrock, Nicholas D, Sederberg, Per B, Ting, Jordyn E, Sharma, Gaurav, Bockbrader, Marcia A, Friedenberg, David A
• Format: Journal Article
• Language: English
• Published: United States Nature Publishing Group 01.11.2018
• Subjects: Adult; Algorithms; Artificial neural networks; Brain - physiopathology; Brain-Computer Interfaces - standards; Brain-Computer Interfaces - trends; Decoders; Decoding; Electric Stimulation; Electrical stimuli; Forearm; Hand Strength - physiology; Human-computer interface; Humans; Male; Motivation - physiology; Movement - physiology; Nerve Net - physiopathology; Neural coding; Neural networks; Paralysis; Quadriplegia - physiopathology; Quadriplegia - rehabilitation; Retraining; User-Computer Interface
• ISSN: 1078-8956; 1546-170X
• DOI: 10.1038/s41591-018-0171-y

Brain-computer interface (BCI) neurotechnology has the potential to reduce disability associated with paralysis by translating neural activity into control of assistive devices . Surveys of potential end-users have identified key BCI system features , including high accuracy, minimal daily setup, rapid response times, and multifunctionality. These performance characteristics are primarily influenced by the BCI's neural decoding algorithm , which is trained to associate neural activation patterns with intended user actions. Here, we introduce a new deep neural network decoding framework for BCI systems enabling discrete movements that addresses these four key performance characteristics. Using intracortical data from a participant with tetraplegia, we provide offline results demonstrating that our decoder is highly accurate, sustains this performance beyond a year without explicit daily retraining by combining it with an unsupervised updating procedure , responds faster than competing methods , and can increase functionality with minimal retraining by using a technique known as transfer learning . We then show that our participant can use the decoder in real-time to reanimate his paralyzed forearm with functional electrical stimulation (FES), enabling accurate manipulation of three objects from the grasp and release test (GRT) . These results demonstrate that deep neural network decoders can advance the clinical translation of BCI technology.