Classification of 4-class Motor Imagery EEG Data with Common Sparse Spectral Spatial Pattern Method

Akinci B., GENÇER N. G.

14th National Biomedical Engineering Meeting, İzmir, Turkey, 20 - 22 May 2009, pp.49-52, (Full Text) identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/biyomut.2009.5130261
  • City: İzmir
  • Country: Turkey
  • Page Numbers: pp.49-52
  • Keywords: Brain Computer Interface, Common Spatial Patterns, EEG, motor imagery, FILTERS
  • Middle East Technical University Affiliated: Yes

Abstract

Brain Computer Interface aims to provide a communication system with external media via thougths. For this purpose, brain signals are acquired from the scalp by EEG device and processed for characterization. In this work, the classification of movement imagery EEG data has been studied for left hand, right hand, foot and tongue movement imagination cases. Common Spatial Patterns (CSP) method and temporal filters have been used in classification and Common Sparse Spectral Spatial Patterns (CSSSP) method has been tried on 4-class motor imagery data in order to improve the accuracy for classification.