Entropy based automatic detection of absence seizures

Seizures can be classified into convulsive and non-convulsive type. First type of seizures, viz., convulsive oneis easy to be diagnosed due to the physical symptoms that are associated with it. Second type of seizures, also called absence seizures is hard to be detected and hence treated. This study focuses on designing of an automatic detection algorithm for absence seizures. Approximate entropy and Energy present in the Delta sub-band of electroencephalogram (EEG), are used as features for classification of normal and absence seizures using a simple linear classifier. When tested on the scalp EEG of two patients having five seizures, the sensitivity and specificity of 98.46% and 99.86% have been obtained by the proposed algorithm. An online detection system can be implemented using the proposed method once it is tested on a large dataset.