Localization and classification of phonemes using high spatial resolution electrocorticography (ECoG) grids

Annu Int Conf IEEE Eng Med Biol Soc. 2008:2008:4964-7. doi: 10.1109/IEMBS.2008.4650328.

Abstract

We present results of cortical activity during phoneme pronunciation, recorded using miniaturized electrocorticography grids with high spatial resolution. A patient implanted with the miniature grid was instructed to audibly pronounce one of four phonemes. For each phoneme, we observed distinct spatial correlation patterns at the 3mm electrode spacing. We applied a support vector machine classification scheme and, for the first time, were able to distinguish discrete phonemes with high accuracy. In addition, we found that sub-regions of our miniature array were specific for distinct pairs of phonemes, showing that cortical phoneme processing occurs at a higher resolution than previously though.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't
  • Research Support, U.S. Gov't, Non-P.H.S.

MeSH terms

  • Algorithms
  • Brain / physiopathology*
  • Brain Mapping / methods*
  • Electrodes, Implanted
  • Electroencephalography / instrumentation
  • Electroencephalography / methods
  • Epilepsy / physiopathology*
  • Female
  • Humans
  • Middle Aged
  • Pattern Recognition, Automated / methods*
  • Phonetics*
  • Sensitivity and Specificity
  • Speech*