Preliminary Comparison of Classifiers to Detect Spatio-spectral Patterns of Epileptic Seizures via PARAFAC Decomposition

Marlis Ontivero-Ortega, Yalina García-Puente, Eduardo Martínez-Montes

2014

Abstract

The automatic detection of epileptic seizures from EEG recording is very important for clinical diagnosis and monitoring and has become an issue of major scientific and technological interest. In this work, we use the spatio-spectral features extracted via multidimensional PARAFAC analysis of the EEG for seizure detection. This is a subject-specific approach which only requires extracting one component explaining a seizure’s space-time-frequency pattern. Then, we propose a simple adaptive zero-training technique (AZT) to classify the seizures, with the additional advantages of being fast and able to be used online. The performance of this technique is evaluated by comparing its accuracy, sensitivity and specificity with those obtained from known pattern recognition methodologies (LDA, SVM, k-Means), on EEG recordings of two epileptic pediatric patients. Results showed that the new method offers the highest sensitivity for different segments’ length, although small segments lead to an increase of the false positives rate. The combination of the feature extracted via PARAFAC model and the AZT procedure would therefore be a promising technique for fast zero-training online seizure detection.

References

1. Miwakeichi F, Martínez-Montes E et al. (2004) Decomposing EEG data into Space-Time-Frequency Components using Parallel Factor Analysis. Neuroimage 22: 1035-1045.
2. Martínez-Montes, E., Márquez-Bocalandro, Y., et al. (2013). EEG Pattern Recognition by Multidimensional

Paper Citation

in Harvard Style

Ontivero-Ortega M., García-Puente Y. and Martínez-Montes E. (2014). Preliminary Comparison of Classifiers to Detect Spatio-spectral Patterns of Epileptic Seizures via PARAFAC Decomposition . In - NEUROTECHNIX, ISBN , pages 0-0

in Bibtex Style

@conference{neurotechnix14,
author={Marlis Ontivero-Ortega and Yalina García-Puente and Eduardo Martínez-Montes},
title={Preliminary Comparison of Classifiers to Detect Spatio-spectral Patterns of Epileptic Seizures via PARAFAC Decomposition },
booktitle={ - NEUROTECHNIX,},
year={2014},
pages={},
publisher={SciTePress},
organization={INSTICC},
doi={},
isbn={},
}

in EndNote Style

TY - CONF
JO - - NEUROTECHNIX,
TI - Preliminary Comparison of Classifiers to Detect Spatio-spectral Patterns of Epileptic Seizures via PARAFAC Decomposition
SN -
AU - Ontivero-Ortega M.
AU - García-Puente Y.
AU - Martínez-Montes E.
PY - 2014
SP - 0
EP - 0
DO -