TY - GEN
T1 - Inter-ictal Seizure Onset Zone localization using unsupervised clustering and Bayesian Filtering
AU - Varatharajah, Yogatheesan
AU - Berry, Brent M.
AU - Kalbarczyk, Zbigniew T.
AU - Brinkmann, Benjamin H.
AU - Worrell, Gregory A.
AU - Iyer, Ravishankar K.
N1 - Funding Information:
X. ACKNOWLEDGMENTS This work was partly supported by National Science Foundation grants CNS-1337732 and CNS-1624790, Na- tional Institute of Health grants NINDS-U01-NS073557, NINDS-R01-NS92882, NHLBI-HL105355, and NINDS-UH2-NS095495-01, Mayo Clinic and Illinois Alliance Fellowships for Technology-based Healthcare Research and IBM faculty award for Ravishankar K. Iyer.
Publisher Copyright:
© 2017 IEEE.
PY - 2017/8/10
Y1 - 2017/8/10
N2 - Surgical removal of seizure-generating brain tissue can cure epilepsy in patients who do not respond to medications. However, identifying seizure-generating regions is difficult and fails in many cases. In this paper, we report a fully unsupervised and automated approach to seizure focus localization using a Bayesian filter. This method uses a spectral domain feature, Power in Bands (PIB). PIB is extracted from inter-ictal (non-seizure) intracranial EEG recordings of patients with focal epilepsy to differentiate normal and abnormal brain regions. This study was carried out using data collected from 34 patients with focal epilepsy at the Mayo Clinic. Experiments show that using a Bayesian filter for capturing temporal properties of the iEEGs recorded from epileptic brains remarkably improves localization accuracy (AUC: 0.63 → 0.72). Our study also reaffirms that high-frequency oscillations and inter-ictal spikes are useful inter-ictal biomarkers of the epileptic brain, and PIB, which could be implemented with relatively low computational burden, performs as well as the standard bio-markers when used in this setting.We conclude that the technique of extracting spectral features from inter-ictal iEEGs and capturing their temporal properties via a Bayesian filter markedly improves our ability to localize seizure onset zones.
AB - Surgical removal of seizure-generating brain tissue can cure epilepsy in patients who do not respond to medications. However, identifying seizure-generating regions is difficult and fails in many cases. In this paper, we report a fully unsupervised and automated approach to seizure focus localization using a Bayesian filter. This method uses a spectral domain feature, Power in Bands (PIB). PIB is extracted from inter-ictal (non-seizure) intracranial EEG recordings of patients with focal epilepsy to differentiate normal and abnormal brain regions. This study was carried out using data collected from 34 patients with focal epilepsy at the Mayo Clinic. Experiments show that using a Bayesian filter for capturing temporal properties of the iEEGs recorded from epileptic brains remarkably improves localization accuracy (AUC: 0.63 → 0.72). Our study also reaffirms that high-frequency oscillations and inter-ictal spikes are useful inter-ictal biomarkers of the epileptic brain, and PIB, which could be implemented with relatively low computational burden, performs as well as the standard bio-markers when used in this setting.We conclude that the technique of extracting spectral features from inter-ictal iEEGs and capturing their temporal properties via a Bayesian filter markedly improves our ability to localize seizure onset zones.
UR - http://www.scopus.com/inward/record.url?scp=85028591892&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85028591892&partnerID=8YFLogxK
U2 - 10.1109/NER.2017.8008407
DO - 10.1109/NER.2017.8008407
M3 - Conference contribution
AN - SCOPUS:85028591892
T3 - International IEEE/EMBS Conference on Neural Engineering, NER
SP - 533
EP - 539
BT - 8th International IEEE EMBS Conference on Neural Engineering, NER 2017
PB - IEEE Computer Society
T2 - 8th International IEEE EMBS Conference on Neural Engineering, NER 2017
Y2 - 25 May 2017 through 28 May 2017
ER -