Classification of patterns of EEG synchronization for seizure prediction

被引:274
作者
Mirowski, Piotr [1 ]
Madhavan, Deepak [2 ]
LeCun, Yann [1 ]
Kuzniecky, Ruben [3 ]
机构
[1] NYU, Courant Inst Math Sci, New York, NY 10003 USA
[2] 982045 Univ Nebraska Med Ctr, Dept Neurol Sci, Omaha, NE 68198 USA
[3] NYU, Comprehens Epilepsy Ctr, New York, NY 10016 USA
关键词
Seizure prediction; Feature extraction; Classification; Pattern recognition; Machine learning; Neural networks; EPILEPTIC SEIZURES; LONG; ANTICIPATION; SELECTION; ENERGY; STATE;
D O I
10.1016/j.clinph.2009.09.002
中图分类号
R74 [神经病学与精神病学];
学科分类号
摘要
Objective: Research in seizure prediction from intracranial EEG has highlighted the usefulness of bivariate measures of brainwave synchronization. Spatio-temporal bivariate features are very high-dimensional and cannot be analyzed with conventional statistical methods. Hence, we propose state-of-the-art machine learning methods that handle high-dimensional inputs. Methods: We computed bivariate features of EEG synchronization (cross-correlation, nonlinear interdependence, dynamical entrainment or wavelet synchrony) on the 21-patient Freiburg dataset. Features from all channel pairs and frequencies were aggregated over consecutive time points, to form patterns. Patient-specific machine learning-based classifiers (support vector machines, logistic regression or convolutional neural networks) were trained to discriminate interictal from preictal patterns of features. In this explorative study, we evaluated out-of-sample seizure prediction performance, and compared each combination of feature type and classifier. Results: Among the evaluated methods, convolutional networks combined with wavelet coherence successfully predicted all out-of-sample seizures, without false alarms, on 15 patients, yielding 71% sensitivity and 0 false positives. Conclusions: Our best machine learning technique applied to spatio-temporal patterns of EEG synchronization outperformed previous seizure prediction methods on the Freiburg dataset. Significance: By learning spatio-temporal dynamics of EEG synchronization, pattern recognition could capture patient-specific seizure precursors. Further investigation on additional datasets should include the seizure prediction horizon. (C) 2009 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
引用
收藏
页码:1927 / 1940
页数:14
相关论文
共 40 条
[1]   Testing the null hypothesis of the nonexistence of a preseizure state [J].
Andrzejak, RG ;
Mormann, F ;
Kreuz, T ;
Rieke, C ;
Kraskov, A ;
Elger, CE ;
Lehnertz, K .
PHYSICAL REVIEW E, 2003, 67 (01) :4
[2]   A robust method for detecting interdependences: application to intracranially recorded EEG [J].
Arnhold, J ;
Grassberger, P ;
Lehnertz, K ;
Elger, CE .
PHYSICA D-NONLINEAR PHENOMENA, 1999, 134 (04) :419-430
[3]   How well can epileptic seizures be predicted? An evaluation of a nonlinear method [J].
Aschenbrenner-Scheibe, R ;
Maiwald, T ;
Winterhalder, M ;
Voss, HU ;
Timmer, J ;
Schulze-Bonhage, A .
BRAIN, 2003, 126 :2616-2626
[4]  
BOTTOU L, 2002, LUSH LISP UNIVERSAL
[5]   LIBSVM: A Library for Support Vector Machines [J].
Chang, Chih-Chung ;
Lin, Chih-Jen .
ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2011, 2 (03)
[6]  
Cherkassky V, 1997, IEEE Trans Neural Netw, V8, P1564, DOI 10.1109/TNN.1997.641482
[7]  
CORTES C, 1995, MACH LEARN, V20, P273, DOI 10.1023/A:1022627411411
[8]   A multi-feature and multi-channel univariate selection process for seizure prediction [J].
D'Alessandro, M ;
Vachtsevanos, G ;
Esteller, R ;
Echauz, J ;
Cranstoun, S ;
Worrell, G ;
Parish, L ;
Litt, B .
CLINICAL NEUROPHYSIOLOGY, 2005, 116 (03) :506-516
[9]   Epileptic seizure prediction using hybrid feature selection over multiple intracranial EEG electrode contacts: A report of four patients [J].
D'Alessandro, M ;
Esteller, R ;
Vachtsevanos, G ;
Hinson, A ;
Echauz, J ;
Litt, B .
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, 2003, 50 (05) :603-615
[10]   EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis [J].
Delorme, A ;
Makeig, S .
JOURNAL OF NEUROSCIENCE METHODS, 2004, 134 (01) :9-21