Investigations of brain -machine interface system-control exploiting local field potential oscillations in motor cortex.

Abstract

Advances in neural recording technologies and electrophysiological techniques have spawned new opportunities to aid in the pursuit of clinical Brain-machine Interface (BMI) systems. Real-time direct interfaces between the brain and electronic or mechanical devices may one day commonly be used to re-establish sensory and motor functions lost through injury or disease bringing new hope to the 250,000 individuals with severe neuromuscular impairments. Previous studies have demonstrated the feasibility of utilizing BMI-systems for neurological augmentation; however many aspects of the system still require further investigation and optimization before being clinically available. The research presented describes investigations of three vital aspects of a BMI-system: (1) characterization of chronically implanted cortical probes, (2) electrophysiological characterization of movement-related Local Field Potentials (LFPs), and (3) investigation of BMI-system-control exploiting LFPs in the motor cortex. These experiments were implemented in the rat model with the following results: (1) Silicon-based Michigan probes reliably provided high-quality spike and LFP recordings over extended periods of time in excess of 1 year. Over 92% of the individual recording sites (n = 224) contained single- or multi-unit activity and nearly 100% contained viable LFP activity throughout this assessment period. (2) Movement-related LFP oscillations were moderately correlated (0.38 +/- 0.14) between the upper and lower cortical layers during non-movement-related epochs and highly correlated (-0.82 +/- 0.18) during stereotypical movement-related epochs. Movement-related synaptic current flow was present in the deeper layers of the motor cortex as well as the superficial layers. This demonstrates some level of communication between the two cortical depths, exclusively during movement. (3) Subjects were consistently able to modulate LFP activity significantly above chance within two days of training. As the subjects learned the task, the intensity of synaptic current flow in the lower cortical layers, the output of motor cortex, diminished significantly, demonstrating that modulation of LFP activity predominantly located in the superficial layers was adequate to accomplish the task. The results from this study demonstrate that these planar silicon probes are suitable for long-term recording in the cerebral cortex and LFPs are ample neural signals to be exploited for BMI-system-control.