QuPWM: Feature Extraction Method for MEG Epileptic Spike Detection
Name:
Journal2019_PWM_Epilepsy_spikes_Detection (2).pdf
Size:
2.819Mb
Format:
PDF
Description:
Preprint
Type
PreprintAuthors
Chahid, Abderrazak
Albalawi, Fahad
Alotaiby, Turky Nayef
Al-Hameed, Majed Hamad
Alshebeili, Saleh
Laleg-Kirati, Taous-Meriem

KAUST Department
Computational Bioscience Research CenterComputational Bioscience Research Center (CBRC)
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Electrical Engineering
Electrical Engineering Program
KAUST Grant Number
BAS/1/1627-01-01Date
2019-07-03Permanent link to this record
http://hdl.handle.net/10754/655961
Metadata
Show full item recordSummary
This record has been merged with an existing record at: http://hdl.handle.net/10754/661489.
Abstract
Epilepsy is a neurological disorder classified as the second most seriousneurological disease known to humanity, after stroke. Localization of theepileptogenic zone is an important step for epileptic patient treatment, whichstarts with epileptic spike detection. The common practice for spike detectionof brain signals is via visual scanning of the recordings, which is asubjective and a very time-consuming task. Motivated by that, this paperfocuses on using machine learning for automatic detection of epileptic spikesin magnetoencephalography (MEG) signals. First, we used the Position WeightMatrix (PWM) method combined with a uniform quantizer to generate usefulfeatures. Second, the extracted features are classified using a Support VectorMachine (SVM) for the purpose of epileptic spikes detection. The proposedtechnique shows great potential in improving the spike detection accuracy andreducing the feature vector size. Specifically, the proposed technique achievedaverage accuracy up to 98\% in using 5-folds cross-validation applied to abalanced dataset of 3104 samples. These samples are extracted from 16 subjectswhere eight are healthy and eight are epileptic subjects using a sliding frameof size of 100 samples-points with a step-size of 2 sample-pointsSponsors
Acknowledgment: Research reported in this publication was supported by King Abdullah University of Science and Technology (KAUST) in collaboration with King Abdulaziz City for Science and Technology (KACST) and King Saud University (KSU). Funding: This research project has been funded by King Abdullah University of Science and Technology (KAUST) Base Research Fund (BAS/1/1627-01-01), in collaboration with King Abdulaziz City for Science and Technology (KACST) and King Saud University (KSU).Publisher
arXivarXiv
1907.02596Additional Links
https://arxiv.org/pdf/1907.02596https://github.com/EMANG-KAUST/QuPWM