Enhancing the detection of seizures with a clustering algorithm

A. Klatchko, G. Raviv, William Webber, Ronald P Lesser

Research output: Contribution to journalArticle

Abstract

Automated detection algorithms of EEG seizures or similar clinical events typically analyze a finite epoch a given channel at a time, producing a probability or a weight estimating how likely it is for the event to resemble a clinical pattern. Epochs are normally shorter than the duration of a seizure, which may spread to more than one electrode. This may result in a weak correspondence between the seizure pattern in the record and its calculated detector event counterpart. As a result, such algorithms suffer from a high rate of false detections. We show that the weights/probabilities of a generic detector can be described as a weight function embedded in a directed graph (digraph). Extended objects such as seizures therefore correspond to the connected components of the digraph. We introduce a clustering algorithm that accounts for the shortcomings of a generic detector of the type described above. By correlating detector results with respect to both time and channel, we effectively extend the detection to an unlimited number of electrodes over an indefinite time. The algorithm is fast (linear - O(m)) and may be implemented in real time. We argue that the algorithm enhances the detection of seizure onset and lowers the rate of false detections. Preliminary results demonstrate a strong correlation between the seizure and the cluster's boundaries and over 50% reduction of false detection rate.

Original languageEnglish (US)
Pages (from-to)52-63
Number of pages12
JournalElectroencephalography and Clinical Neurophysiology
Volume106
Issue number1
DOIs
StatePublished - Jan 1998

Fingerprint

Cluster Analysis
Seizures
Weights and Measures
Electrodes
Electroencephalography

Keywords

  • Clustering algorithms
  • EEG
  • Graph theory
  • Seizure detection

ASJC Scopus subject areas

  • Clinical Neurology
  • Neuroscience(all)

Cite this

Enhancing the detection of seizures with a clustering algorithm. / Klatchko, A.; Raviv, G.; Webber, William; Lesser, Ronald P.

In: Electroencephalography and Clinical Neurophysiology, Vol. 106, No. 1, 01.1998, p. 52-63.

Research output: Contribution to journalArticle

@article{ecc7cea13e7e40c6b07d5dcf5d833548,
title = "Enhancing the detection of seizures with a clustering algorithm",
abstract = "Automated detection algorithms of EEG seizures or similar clinical events typically analyze a finite epoch a given channel at a time, producing a probability or a weight estimating how likely it is for the event to resemble a clinical pattern. Epochs are normally shorter than the duration of a seizure, which may spread to more than one electrode. This may result in a weak correspondence between the seizure pattern in the record and its calculated detector event counterpart. As a result, such algorithms suffer from a high rate of false detections. We show that the weights/probabilities of a generic detector can be described as a weight function embedded in a directed graph (digraph). Extended objects such as seizures therefore correspond to the connected components of the digraph. We introduce a clustering algorithm that accounts for the shortcomings of a generic detector of the type described above. By correlating detector results with respect to both time and channel, we effectively extend the detection to an unlimited number of electrodes over an indefinite time. The algorithm is fast (linear - O(m)) and may be implemented in real time. We argue that the algorithm enhances the detection of seizure onset and lowers the rate of false detections. Preliminary results demonstrate a strong correlation between the seizure and the cluster's boundaries and over 50{\%} reduction of false detection rate.",
keywords = "Clustering algorithms, EEG, Graph theory, Seizure detection",
author = "A. Klatchko and G. Raviv and William Webber and Lesser, {Ronald P}",
year = "1998",
month = "1",
doi = "10.1016/S0013-4694(97)00092-8",
language = "English (US)",
volume = "106",
pages = "52--63",
journal = "Electroencephalography and Clinical Neurophysiology",
issn = "0013-4694",
publisher = "Elsevier BV",
number = "1",

}

TY - JOUR

T1 - Enhancing the detection of seizures with a clustering algorithm

AU - Klatchko, A.

AU - Raviv, G.

AU - Webber, William

AU - Lesser, Ronald P

PY - 1998/1

Y1 - 1998/1

N2 - Automated detection algorithms of EEG seizures or similar clinical events typically analyze a finite epoch a given channel at a time, producing a probability or a weight estimating how likely it is for the event to resemble a clinical pattern. Epochs are normally shorter than the duration of a seizure, which may spread to more than one electrode. This may result in a weak correspondence between the seizure pattern in the record and its calculated detector event counterpart. As a result, such algorithms suffer from a high rate of false detections. We show that the weights/probabilities of a generic detector can be described as a weight function embedded in a directed graph (digraph). Extended objects such as seizures therefore correspond to the connected components of the digraph. We introduce a clustering algorithm that accounts for the shortcomings of a generic detector of the type described above. By correlating detector results with respect to both time and channel, we effectively extend the detection to an unlimited number of electrodes over an indefinite time. The algorithm is fast (linear - O(m)) and may be implemented in real time. We argue that the algorithm enhances the detection of seizure onset and lowers the rate of false detections. Preliminary results demonstrate a strong correlation between the seizure and the cluster's boundaries and over 50% reduction of false detection rate.

AB - Automated detection algorithms of EEG seizures or similar clinical events typically analyze a finite epoch a given channel at a time, producing a probability or a weight estimating how likely it is for the event to resemble a clinical pattern. Epochs are normally shorter than the duration of a seizure, which may spread to more than one electrode. This may result in a weak correspondence between the seizure pattern in the record and its calculated detector event counterpart. As a result, such algorithms suffer from a high rate of false detections. We show that the weights/probabilities of a generic detector can be described as a weight function embedded in a directed graph (digraph). Extended objects such as seizures therefore correspond to the connected components of the digraph. We introduce a clustering algorithm that accounts for the shortcomings of a generic detector of the type described above. By correlating detector results with respect to both time and channel, we effectively extend the detection to an unlimited number of electrodes over an indefinite time. The algorithm is fast (linear - O(m)) and may be implemented in real time. We argue that the algorithm enhances the detection of seizure onset and lowers the rate of false detections. Preliminary results demonstrate a strong correlation between the seizure and the cluster's boundaries and over 50% reduction of false detection rate.

KW - Clustering algorithms

KW - EEG

KW - Graph theory

KW - Seizure detection

UR - http://www.scopus.com/inward/record.url?scp=0031888476&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031888476&partnerID=8YFLogxK

U2 - 10.1016/S0013-4694(97)00092-8

DO - 10.1016/S0013-4694(97)00092-8

M3 - Article

VL - 106

SP - 52

EP - 63

JO - Electroencephalography and Clinical Neurophysiology

JF - Electroencephalography and Clinical Neurophysiology

SN - 0013-4694

IS - 1

ER -