A New Approach to Automatic Epilepsy Detection from EEG Signals Using Archimedean Spiral and Swin Transformer

Epilepsy, a neurological disorder marked by recurrent and unpredictable seizures due to abnormal brain electrical activity, is studied using electroencephalography (EEG) which measures brain activity. The EEG signals are commonly employed to diagnose and monitor conditions like epilepsy, sleep disorders, and brain injuries. This research work introduces an effective hybrid approach based on Archimedean spiral coding (ASC) and a swin transformer-based convolutional neural network (CNN) techniques to detect epilepsy automatically using EEG signals. This proposed ASC method transforms EEG signals into a visually informative 3D matrix and employs swin transformer-based CNN architecture for classification. It yields an accuracy of 97.98% and 88.22% for sample- and subject-based ten-fold cross-validation, respectively using the public epilepsy database of 121 populations. The developed system is ready to be tested with more epilepsy patients from different races to validate the performance. It produced a better accuracy score as compared to the existing results.