Single-Trial Classification of MEG Recordings

• Published in: IEEE transactions on biomedical engineering Vol. 54; no. 3; pp. 436 - 443
• Main Authors: Perreau Guimaraes, Marcos, Wong, Dik Kin, Uy, E. Timothy, Grosenick, Logan, Suppes, Patrick
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.03.2007; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Brain; Brain Mapping - methods; classification; Cluster Analysis; Diagnosis, Computer-Assisted - methods; Electroencephalography; Evoked Potentials, Auditory - physiology; Evoked Potentials, Visual - physiology; Humans; ICA; Independent component analysis; linear discriminant; Linear discriminant analysis; Magnetic fields; Magnetic noise; Magnetic sensors; Magnetoencephalography; Magnetoencephalography - methods; MEG; Pattern Recognition, Automated - methods; PCA; Principal Component Analysis; Principal components analysis; single trial; SOBI; Speech Perception - physiology; Studies; SVM; Vectors
• ISSN: 0018-9294; 1558-2531
• DOI: 10.1109/TBME.2006.888824

While magnetoencephalography (MEG) is widely used to identify spatial locations of brain activations associated with various tasks, classification of single trials in stimulus-locked experiments remains an open subject. Very significant single-trial classification results have been published using electroencephalogram (EEG) data, but in the MEG case, the weakness of the magnetic fields originating from the relevant sources relative to external noise, and the high dimensionality of the data are difficult obstacles to overcome. We present here very significant MEG single-trial mean classification rates of words. The number of words classified varied from seven to nine and both visual and auditory modalities were studied. These results were obtained by using a variety of blind sources separation methods: spatial principal components analysis (PCA), Infomax independent components analysis (Infomax ICA) and second-order blind identification (SOBI). The sources obtained were classified using two methods, linear discriminant classification (LDC) and nu-support vector machine (nu-SVM). The data used here, auditory and visual presentations of words, presented nontrivial classification problems, but with Infomax ICA associated with LDC we obtained high classification rates. Our best single-trial mean classification rate was 60.1% for classification of 900 single trials of nine auditory words. On two-class problems rates were as high as 97.5%