Dergiler / Avrupa Bilim ve Teknoloji Dergisi / 2022 / Sayı: 35
Analyzing of EEG Signals with Deep Learning and Discrete Wavelet Transform
- Sayı
- 2022 · Sayı: 35
- Sayfa
- 514–521
- DOI
- —
Özet
In this study, the capability to study the effect of each feature on the accuracy of the classification, whereby in the mixture of features with the Convolutional Neural Networks (CNNs) to identify epilepsy seizure in EEGs was searched. The EEG signals were first analyzed within 5 subsignals at specific frequencies bands by using Discrete Wavelet Transforms (DWT) at 5 levels, and then features were extracted from each sub signal. Finally, there was convolutional neural network classification. The best classification accuracies obtained when extracted eight features from EEG signals 96.5%. That means these features are strong to catch epilepsy seizure. Usually, the smart methods could be utilized within a more broad range of identification model problems that are also relevant to humans, such as the epilepsy diseases discovery and judgment.
Abstract
In this study, the capability to study the effect of each feature on the accuracy of the classification, whereby in the mixture of features with the Convolutional Neural Networks (CNNs) to identify epilepsy seizure in EEGs was searched. The EEG signals were first analyzed within 5 subsignals at specific frequencies bands by using Discrete Wavelet Transforms (DWT) at 5 levels, and then features were extracted from each sub signal. Finally, there was convolutional neural network classification. The best classification accuracies obtained when extracted eight features from EEG signals 96.5%. That means these features are strong to catch epilepsy seizure. Usually, the smart methods could be utilized within a more broad range of identification model problems that are also relevant to humans, such as the epilepsy diseases discovery and judgment.