Work

構造化信号表現とドメイン知識を用いた長時間eegのための機械学習によるてんかん発作検出

Year: 2025
Type: dissertation
Abstract: The scalp electroencephalogram (EEG) is a non-invasive method to record brain electrical activity that has been used extensively in epilepsy diagnosis.Manual seizure event detection from long EEG recordings is a cumbersome process, making automated seizure detection increasingly important for practitioners.Automated seizure detection faces several challenges, including feature engineering, inter-patient variability, and artifacts.However, most existing approaches struggle to accurately distinguish seizure EEG from artifacts or noise, which are commonly present in scalp EEG recordings.This thesis addresses those challenges through two complementary approaches that make use of structured signal representations and domain knowledge of EEG signals.The first approach is a multi-stage pipeline that integrates anomaly detection, artifact rejection, and seizure classifica-First and foremost, I would like to express my deepest gratitude to my supervisor, Professor Dr. Toshihisa Tanaka, for his invaluable guidance, continuous support, and insightful feedback throughout the course of my research
Language: English
Open Access status: green
OpenAlex ID: W7154548072
Cites: 114
Cited by:
Related to: 10