Epileptic seizures detection is largely based on analysis of Electroencephalogram signals. The ambulatory EEG recordings generate very lengthy data which require a skilled and careful analysis. This tedious procedure necessitates the use of automated systems for epileptic seizure detection. This paper proposes one such automated epileptic seizure detection technique based on Probabilistic Neural Network (PNN) by using a time frequency domain characteristics of EEG signal called Approximate Entropy (ApEn). Our method consists of EEG data collection, feature extraction and classification. EEG data from normal and epileptic subjects was collected, digitized and then fed into the PNN. For feature extraction, the wavelet coefficients are derived using Discrete Wavelet Transformation. For the feature selection stage a new methodology is proposed, which is, comparing the ApEn values of wavelet coefficients of different EEG data. The experimental results portray that this proposed approach efficiently detects the presence of epileptic seizures in EEG signals and showed a reasonable accuracy.
Sharma, Garima and Pradhan, N.
"Implementation of Probabilistic Neural Network using Approximate Entropy to Detect Epileptic Seizures,"
Undergraduate Academic Research Journal: Vol. 1:
1, Article 4.
Available at: https://www.interscience.in/uarj/vol1/iss1/4