Abstract— Epilepsy is a neurological disorder that affects around 50 million people worldwide. The seizure detection is an important component in the diagnosis of epilepsy. In this study, the Empirical Mode Decomposition (EMD) method was proposed on the development of an automatic epileptic seizure detection algorithm. The algorithm first computes the Intrinsic Mode Functions (IMFs) of EEG records, then calculates the energy of each IMF and performs the detection based on an energy threshold and a minimum duration decision. The algorithm was tested in 9 invasive EEG records provided and validated by the Epilepsy Center of the University Hospital of Freiburg. In 90 segments analyzed (39 with epileptic seizures) the sensitivity and specificity obtained with the method were of 56.41% and 75.86% respectively. It could be concluded that EMD is a promissory method for epileptic seizure detection in EEG records. I. INTRODUCTION PILEPSY is a chronic neurological disorder that affects around 50 million people worldwide of all ages [1]. This brain disorder is characterized by recurrent seizures which are the clinical manifestations of sudden, usually brief, excessive electrical discharges in a group of brain cells. Different parts of the brain can be the source of such discharges. Epilepsy responds to antiepileptic drugs about 70% of the cases and the remaining affected individuals could benefit from surgical therapy [1]. The seizure detection is an important component in the diagnosis of epilepsy. This includes visual scanning of Electroencephalogram (EEG) long recordings which is very time consuming and the conclusions are very subjective so disagreement between physicians are not rare. For this reason, the computerized analysis of EEG signals using automatic algorithms is highly useful for the diagnosis of this disease. Several individual processing techniques and also a This work has been supported by grants from Agencia Nacional de Promoción Científica y Tecnológica (PICT 2006-01689) and Universidad Nacional de San Juan, both institutions from Argentina, and by Ministerio de Ciencia e Innovación de España (TEC2007-68076-C02-01). The first author is supported by ANCYT, whereas the second and third authors are supported by CONICET of Argentina. L. Orosco, E. Laciar, A. Garcés and J. P. Graffigna are with Gabinete de Tecnología Médica,Universidad Nacional de San Juan, San Juan, Argentina (e-mail: lorosco@gateme.unsj.edu.ar, laciar@gateme.unsj.edu.ar, agarces@gateme.unsj.edu.ar, jgraffig@gateme.unsj.edu.ar). A. Torres is with Dept. ESAII, Universitat Politècnica de Catalunya, Institut de Bioenginyeria de Catalunya (IBEC) and CIBER de Bioingeniería, Biomateriales y Nanomedicina (CIBER-BBN), Barcelona, España. E-mail: abel.torres@upc.edu. combination of those were employed and refined for epileptic seizures detection, quantification and recognition [2]. Neural Networks (NN) have been used to detect abnormal patterns in the EEG [3] and to identify seizure or preseizure states [4]. Wavelet Transform is also widely used for epilepsy detection [4], [5]. Others studies combine Approximate Entropy and Lempel-Ziv Complexity [6], and Time Frequency Distributions and NN [7]. In the last years, a new technique called empirical mode decomposition (EMD) has been proposed for the analysis of non-linear and non-stationary series. The EMD technique decomposed a series set into a finite and often small number of intrinsic mode functions (IMF) that admit well-behaved Hilbert transforms [8]. In the field of biomedical signal processing, EMD has been used for the analysis of respiratory mechanomyographic signals [9], for denoising in ECG records [10], and for tracking alpha rhythm in EEG recordings [11] and recently for epileptic seizure detection in EEG signals [12]. In this paper an epileptic seizure detection method based on EMD of EEG signal is proposed. The method computes the different IMFs of the signals and the seizure is detected applying energy and duration criteria on these IMFs. II. MATERIALS The EEG database contains invasive EEG recordings of 21 patients suffering from medically intractable focal epilepsy. The data were recorded during invasive pre- surgical epilepsy monitoring at the Epilepsy Center of the University Hospital of Freiburg, Germany [13]. In order to obtain a high signal-to-noise ratio, fewer artifacts, and to record directly from focal areas, intracranial grid-, strip-, and depth-electrodes were used. The EEG data were acquired using a Neurofile NT digital video EEG system with 128 channels, 256 Hz sampling rate, and a 16 bits A/D converter. Notch or band pass filters have not been applied in the acquisition stage. The available EEG records include only 6 channels (3 focal electrodes and 3 extrafocal electrodes). The records are divided into segments of 1hour long. In this study, the 3 intra source records of 9 patients with focal epilepsy originated in the temporal lobe region were selected. This computes a total of 90 segments per each channel, 51 of them without epileptic seizures and 39 segments denoted as having only one epileptic seizure each. An Epileptic Seizures Detection Algorithm based on the Empirical Mode Decomposition of EEG Lorena Orosco, Eric Laciar, Member, IEEE, Agustina Garcés Correa, Abel Torres, Member, IEEE, and Juan P. Graffigna, Member, IEEE E 2651 31st Annual International Conference of the IEEE EMBS Minneapolis, Minnesota, USA, September 2-6, 2009 978-1-4244-3296-7/09/$25.00 ©2009 IEEE