Picture of automobile manufacturing plant

Driving innovations in manufacturing: Open Access research from DMEM

Strathprints makes available Open Access scholarly outputs by Strathclyde's Department of Design, Manufacture & Engineering Management (DMEM).

Centred on the vision of 'Delivering Total Engineering', DMEM is a centre for excellence in the processes, systems and technologies needed to support and enable engineering from concept to remanufacture. From user-centred design to sustainable design, from manufacturing operations to remanufacturing, from advanced materials research to systems engineering.

Explore Open Access research by DMEM...

Extracting multi-dimensional control signals from self-initiated movements

Conway, B.A. and Valsan, G. and Lakany, H. (2008) Extracting multi-dimensional control signals from self-initiated movements. In: Neuroscience 2008, 2008-11-15 - 2008-11-19.

Full text not available in this repository. Request a copy from the Strathclyde author

Abstract

A Brain computer interface (BCI) is a prosthetic system which augments the capabilities of an impaired nervous system thereby allowing a paralyzed individual to regain ability in communication and device control. The aim of this research is to extract multi-dimensional control signals for BCI by distinguishing between different directions of self-initiated wrist movements. Experiments were performed with local ethical approval on 4 subjects who gave their informed consent. The subjects were seated in front of a large computer monitor and held a manipulandum in their right hand. The manipulandum translated the wrist movements of the subject into a cursor movement on the monitor. 28 EEG channels were recorded (Neuroscan Synamp) from sintered Ag/AgCl electrodes located on the scalp using an Easycap in a high density montage covering the wrist and arm areas of the left motor cortex. Four bipolar EMG channels from selected wrist muscles were also recorded. The EEG and EMG signals were digitised as were position and kinematic data derived from the manipulandum which were used to determine movement onset times. In these experiments the subjects were given a free choice of when to initiate wrist movements in any of four directions (up, down, left or right) in a centre movement task. A grid on the PC monitor was used to illustrate the chosen motion once it had been initiated. For each subject at least 100 repeat movements for each direction were acquired. EEG epochs were generated based on the time of movement initiation and analyzed off-line. Principal component analysis of the pre-movement EEG spectrogram was used for feature extraction. Selected features were classified based on a Euclidean distance based classifier to predict direction of the movements. High classification rates were achieved that are comparable or better than achieved with cued movement trials. Classification success varied between subjects as did the electrode sites that generated the best rates. We conclude that EEG signals preceding self-initiated movements associated have the potential to be utilized in development of control signals for BCIs.