114 Enhancement of Cortical Signals to Improve Brain Machine Interface Performance by Inducing Mirror Neuron Network Activation

• Published in: Neurosurgery Vol. 62; no. CN_suppl_1; pp. 201 - 202
• Main Authors: Lobel, Darlene Angela, Achey, Rebecca, Brennan, Ryan, Rastogi, Anisha, Ajiboye, Abidemi B.
• Format: Journal Article
• Language: English
• Published: Philadelphia Oxford University Press 01.08.2015; Copyright by the Congress of Neurological Surgeons; Wolters Kluwer Health, Inc
• Subjects: Discriminant analysis; Neurosurgery; Virtual reality
• ISSN: 0148-396X; 1524-4040
• DOI: 10.1227/01.neu.0000467076.17237.fd

Abstract INTRODUCTION: Brain Machine Interface (BMI) systems allow patients with neurological injuries to control assistive devices using cortical signals produced during motor imagery (MI). To improve BMI performance, we sought to enhance cortical signals by training subjects in MI techniques that induce activation of mirror neuron networks (MNN). METHODS: Four subjects were trained to perform MI of three tasks by either observing an avatar perform movements in a virtual reality (VR) setting, or by watching a video of a researcher performing the tasks (action observation—AO). Baseline MI ability was assessed using a customized scale. Mirror neuron network activation patterns were evaluated in healthy subjects using EEG at baseline, during an MI training session, and after 12 training sessions. Cortical signals were processed using common average referencing across various frequency bands for each of the recording electrodes. Task classification was performed using linear discriminant analysis. Two-dimensional head plots were created to display quantification and spatial distribution of neural activation patterns during performance and imagination of the 3 tasks. RESULTS: Cortical activations during MI and performance of movement increased in intensity across all frequency bands between 100% and 300% of baseline after MI training in all subjects, regardless of training paradigm. This effect was most pronounced in beta (12–30 Hz) and high gamma (50–70 Hz) frequencies. Increased activations were spatially located in motor, premotor, and supplementary motor areas and were noted in both contralateral and ipsilateral cortices in both VR subjects and 1 AO subject. CONCLUSION: Motor imagery training that activates mirror neuron networks enhances cortical signals during MI and during the performance of corresponding movements in healthy subjects. As MI-based signals are used to operate BMIs, our research suggests that MNN-based MI training may improve BMI performance. Additionally, VR-based imagery training may provide a benefit over AO training to induce bilateral activation of MNNs.