Phase space analysis of Event Related Potential during episodic memory retrieval

M Ghoshuni; A M Nasrabadi; M A Khalilzadeh; M R Hashemi

doi:10.1109/IEMBS.2007.4352710

Phase space analysis of Event Related Potential during episodic memory retrieval

Annu Int Conf IEEE Eng Med Biol Soc. 2007:2007:1996-9. doi: 10.1109/IEMBS.2007.4352710.

Authors

M Ghoshuni¹, A M Nasrabadi, M A Khalilzadeh, M R Hashemi

Affiliation

¹ Department of Biomedical Engineering, Islamic Azad University, Mashhad Branch, Mashhad, Iran. ghoshuni@gmail.com

PMID: 18002376
DOI: 10.1109/IEMBS.2007.4352710

Abstract

In this study we have used recurrence analysis (RA) to classify ERPs that appear from episodic memory retrieval in old/new recognition task. Since RA is based on embedding phase space, we have used correlation dimension and autocorrelation function in order to estimate embedding dimension and the lag time between successive components of each of embedding space vectors alternatively. According to RA the rates of classification have been improved in comparison to previous study. We have obtained 98.9% accuracy for train data and 97.7% for test data. Furthermore we could classify ERPs with line interference noise using RA.

MeSH terms

Algorithms
Data Interpretation, Statistical*
Electroencephalography / instrumentation*
Equipment Design
Evoked Potentials*
Humans
Male
Memory*
Mental Processes
Models, Statistical
Models, Theoretical
Reaction Time
Reproducibility of Results
Signal Processing, Computer-Assisted*