Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/4828
Title: Automated Epileptic Seizure Detection from Electroencephalogram Signals
Authors: Jindal, Komal
Upadhyay, Rahul (Guide)
Keywords: EEG
Epilipsy Detection
Machine Learning Technique
Waelet Transform
Issue Date: 5-Sep-2017
Abstract: iii ABSTARCT Epileptic seizure is one of the brain’s disorder which can be automatically diagnosed by measuring and analys ing the non - linear and non - stationary behaviour of brain electrical activity. It is a transient symptom of excessive or synchronous neuronal activity of human brain. It is a group of disorders of human brain which has affected a large part of world’s popul ation. Epileptic seizure disturbs usual pattern of neuronal activity that causes interruption of consciousness, weird sensation and muscle fits. Early recognition of epileptic seizure helps in improving the physiological condition of patient. Human brain e lectrical activity varies with various physiological and neurological conditions and is recorded by multiple scalp mounted electrodes. The record of human brain electrical activity is called Electroencephalogram (EEG) signals. The Electroencephalogram (EEG ) signals are employed to diagnose various human brain disorders. The Electroencephalogram (EEG) signals contain necessary information for early diagnosis of epilepsy and epileptic seizures. In epilepsy, the nerve cells send out high amplitude electrical impulses and the impulses generate events called seizures. In the past, these EEG signals were diagnosed for any brain disorders by visual examination. However, visual examination is susceptible to errors and requires good understanding of EEG activity. T his research work presents an autonomous system, which is capable of detecting epileptic seizure from EEG signals automatically. The proposed system is carried out in three methodological steps viz. pre - processing, feature extraction and classification. T he purpose of pre - processing is to organize the data in an orderly manner and to remove noise. Whereas , feature extraction step extracts time - spectral features for proper representation of seizure and non - seizure signals. Further, the extracted features are then fed to the machine learning algorithms for detection of seizure and non - seizure EEG signals. The proposed system of automatic seizure detection is validated on publicly available dataset and the results show high detection ability of the proposed system. In present work, different feature extraction techniques have been employed and analysed for efficient classification. A comparative study for proposed feature extraction methods is performed in terms of classification efficiency. In this work, Sup port Vector Machine and Artificial Neural Network classifier have been used for classification of Electroencephalogram signals associated with different physiological condition.
Description: Master of Engineering -ECE
URI: http://hdl.handle.net/10266/4828
Appears in Collections:Masters Theses@ECED

Files in This Item:
File Description SizeFormat 
801561012_Komal Jindal_ECED.pdf2.98 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.