Epileptic Seizure State Classification from Scalp EEG using Self-Organizing Maps

Epileptic seizures afflict around one percent of the worldwide human population, are typified by excessive neuronal firing in the human brain, and have a significant influence on the quality of life of the patients. This study demonstrates a Self Organizing Map-based architecture for ternary classification of seizure states into ictal, pre-ictal, and inter-ictal states. The standardized annotated dataset comprising EEG recordings of 24 pediatric individuals collected at Children’s Hospital Boston (CHB-MIT) was used to train the model and evaluate its performance. The suggested model achieved a mean accuracy of 95.63 percent. The model was shown to be similar to advanced classification models on the benchmark dataset, with mean sensitivity rate of 93.51 percent.