Return to search

Detection of Movement Intention Onset for Brain-machine Interfaces

The goal of the study was to use electrical signals from primary motor cortex to generate
accurate predictions of the movement onset time of performed movements, for potential
use in asynchronous brain-machine interface (BMI) systems. Four subjects, two with
electroencephalogram and two with electrocorticogram electrodes, performed various movements while activity from their primary motor cortices was recorded. An analysis program used several criteria (change point, fractal dimension, spectral entropy, sum of differences, bandpower, bandpower integral, phase, and variance), derived from the neural recordings, to generate predictions of movement onset time, which it compared to electromyogram activity onset time, determining prediction accuracy by receiver operating characteristic curve areas. All criteria, excepting phase and change-point analysis, generated accurate predictions in some cases.

Identiferoai:union.ndltd.org:TORONTO/oai:tspace.library.utoronto.ca:1807/18917
Date15 February 2010
CreatorsMcGie, Steven
ContributorsPopovic, Milos R.
Source SetsUniversity of Toronto
Languageen_ca
Detected LanguageEnglish
TypeThesis

Page generated in 0.0016 seconds