Mapping Sensorimotor Function and Controlling Upper Limb Neuroprosthetics with Electrocorticography
Embargo until
Date
2015-09-29
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Johns Hopkins University
Abstract
Electrocorticography (ECoG) occupies a unique intermediate niche between microelectrode recordings of single neurons and recordings of whole brain activity via functional magnetic resonance imaging (fMRI). ECoG’s combination of high temporal resolution and wide area coverage make it an ideal modality for both functional brain mapping and brain-machine interface (BMI) for control of prosthetic devices. This thesis demonstrates the utility of ECoG, particularly in high gamma frequencies (70-120 Hz), for passive online mapping of language and motor behaviors, online control of reaching and grasping of an advanced robotic upper limb, and mapping somatosensory digit representations in the postcentral gyrus. The dissertation begins with a brief discussion of the framework for neuroprosthetic control developed by the collaboration between Johns Hopkins and JHU Applied Physics Laboratory (JHU/APL). Second, the methodology behind an online spatial-temporal functional mapping (STFM) system is described. Trial-averaged spatiotemporal maps of high gamma activity were computed during a visual naming and a word reading task. The system output is subsequently shown and compared to stimulation mapping. Third, simultaneous and independent ECoG-based control of reaching and grasping is demonstrated with the Modular Prosthetic Limb (MPL). The STFM system was used to identify channels whose high gamma power significantly and selectively increases during either reaching or grasping. Using this technique, two patients were able to rapidly achieve naturalistic control over simple movements by the MPL. Next, high-density ECoG (hdECoG) was used to map the cortical responses to mechanical vibration of the fingertips. High gamma responses exhibited a strong yet overlapping somatotopy that was not well replicated in other frequency bands. These responses are strong enough to be detected in single trials and used to classify the finger being stimulated with over 98% accuracy. Finally, the role of ECoG is discussed for functional mapping and BMI applications. ECoG occupies a unique role among neural recording modalities as a tool for functional mapping, but must prove its value relative to stimulation mapping. For BMI, ECoG lags microelectrode arrays but hdECoG may provide a more robust long-term interface with optimal spacing for sampling relevant cortical representations.
Description
Keywords
electrocorticography, brain-machine interface, functional mapping, neuroprosthesis