Exploring dimensionality reduction of EEG features in motor imagery task classification

• Published in: Expert systems with applications Vol. 41; no. 11; pp. 5285 - 5295
• Main Authors: García-Laencina, Pedro J., Rodríguez-Bermudez, Germán, Roca-Dorda, Joaquín
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Ltd 01.09.2014; Elsevier
• Subjects: Applied sciences; Biological and medical sciences; Brain-computer interfaces; Classification; Cognition. Intelligence; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Data processing. List processing. Character string processing; Dimensionality reduction; Electrodiagnosis. Electric activity recording; Electroencephalogram signals; Electroencephalography; Exact sciences and technology; Feature extraction; Feature transformation; Fundamental and applied biological sciences. Psychology; Human-computer interface; Imagery; Investigative techniques, diagnostic techniques (general aspects); Linear discriminants; Local Fisher Discriminant Analysis; Medical sciences; Memory organisation. Data processing; Mental imagery. Mental representation; Motor imagery; Motors; Nervous system; Psychology. Psychoanalysis. Psychiatry; Psychology. Psychophysiology; Reduction; Software; Tasks; Dimensionality reduction; Feature transformation; Local Fisher Discriminant Analysis; Electroencephalogram signals; Brain-computer interfaces; Motor imagery; Linear discriminants; Brain; Electrical activity; Adaptive system; Central nervous system; Electroencephalography; Encephalon; Projection method; User interface; Sequential method; Classification; Selection criterion; Reduced order systems; Discriminant analysis; Computational complexity; Standards; Data reduction; Fisher information; Dimension reduction; Reduction method; Imagination; Intention; User behavior; Principal component analysis
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2014.02.043

•This work analyzes feature selection and transformation methods for BCI systems.•Three representative feature extraction methods are used: BP, Hjorth and AAR.•An efficient LOO-CV technique is introduced for choosing the embedded dimensionality.•Experiments have been conducted on five novice users during their first BCI sessions.•According to its excellent results, LFDA is a promising method to design BCI systems. A Brain-Computer Interface (BCI) system based on motor imagery (MI) identifies patterns of electrical brain activity to predict the user intention while certain movement imagination tasks are performed. Currently, one of the most important challenges is the adaptive design of a BCI system. For solving it, this work explores dimensionality reduction techniques: once features have been extracted from Electroencephalogram (EEG) signals, the high-dimensional EEG data has to be mapped onto a new reduced feature space to make easier the classification stage. Besides the standard sequential feature selection methods, this paper analyzes two unsupervised transformation-based approaches – Principal Component Analysis and Locality Preserving Projections – and the Local Fisher Discriminant Analysis (LFDA), which works in a supervised manner. The dimensionality in the projected space is chosen following a wrapper-based approach by an efficient leave-one-out estimation. Experiments have been conducted on five novice subjects during their first sessions with MI-based BCI systems in order to show that the appropriate use of dimensionality reduction methods allows increasing the performance. In particular, obtained results show that LFDA gives a significant enhancement in classification terms without increasing the computational complexity and, then, it is a promising technique for designing MI-based BCI system.