Classificação de estados epilépticos em sinais de EEG utilizando detecção de anomalias

Abstract

Epilepsy is a neurological disorder characterized by an abnormal electrical disturbance in the brain, causing recurrent seizures. The most commonly used exam to diagnose epilepsy it the electroencephalogram (EEG), where a patient’s brain electrical activity is measured and visually analyzed. However, identifying epileptic patterns in the EEG signal through visual inspection is a time-consuming and exhaustive task for professionals in the field, motivating the development of algorithms that can identify these patterns, aiding the medical diagnosis. In this work, we propose three models based on anomaly detection. The results obtained demonstrate high performance and noise robustness in relation to results found in the literature.