Adaptive decoding using local field potentials in a brain-machine interface.

Brain-machine interface (BMI) systems have the potential to restore function to people who suffer from paralysis due to a spinal cord injury. However, in order to achieve long-term use, BMI systems have to overcome two challenges - signal degeneration over time, and non-stationarity of signals. Effects of loss in spike signals over time can be mitigated by using local field potential (LFP) signals for decoding, and a solution to address the signal non-stationarity is to use adaptive methods for periodic recalibration of the decoding model. We implemented a BMI system in a nonhuman primate model that allows brain-controlled movement of a robotic platform. Using this system, we showed that LFP signals alone can be used for decoding in a closed-loop brain-controlled BMI. Further, we performed offline analysis to assess the potential implementation of an adaptive decoding method that does not presume knowledge of the target location. Our results show that with periodic signal and channel selection adaptation, decoding accuracy using LFP alone can be improved by between 5-50%. These results demonstrate the feasibility of implementing unsupervised adaptive methods during asynchronous decoding of LFP signals for long-term usage in a BMI system.