Unsupervised Artifact Detection From ECoG(iEEG) Signals Using DBSCAN

Artifact detection is an essential step in the preprocessing of electrophysiological signals such as ECoG, as artifacts can significantly distort analysis and lead to misleading conclusions. Traditionally, this task is performed manually by neuroscientists, which is both time-consuming and impractical for large-scale datasets. This work introduces an unsupervised and automatic method for artifact detection using the DBSCAN clustering algorithm. The approach is based on ۱۱ extracted features and does not require any labeled data or human supervision. Despite its simplicity, the method has shown strong performance in identifying artifacts across recordings, as confirmed through visual inspection. This makes it a practical solution for efficient and scalable neural data preprocessing.

کلیدواژه ها:

ECoG ، Artifact Detection ، Machine Learning