A measure of distance for EEG analysis

  • Miguel Ángel Ré CIII - FRC - UTN
  • Juan Manuel Díaz
  • Guillermo Gabriel Aguirre Varela

Abstract

It is presented a quantitative method of analysis of EEG records. The method is based on Jensen-Shannonbdivergence (JSD). The record is segmented and each segment is compared with a reference one through JSD. To calculate JSD, probability distributions are estimated from the values in the EEG record by means of the kernel density approximation. Two cases of study are analyzed and the obtained results are included. The proposed methods allows to distinguish diverse physiological changes, artefacts, nonphysiological artifacts or pathological elements in the EEG record.
Published
2018-08-30
Section
Scientific articles