CLUSTERING AND HYPER LINKING OF LONG TERM EEGS

点击翻译按钮获取中文摘要

DESCRIPTION (Adapted from applicant's abstract): The proposed EEG MagicMarker software offers a new methodology for the display and analysis of digitized EEG records. Segments of similar EEG activity are clustered together, clearly differentiating background, paroxysmal activity, and patient state transitions, e.g., sleep stages. The analysis only considers the content of the current EEG and requires no thresholds or classification functions derived from a training set. A novel user interface allows interactive partitioning of the hierarchical cluster dendrogram in a method already familiar to many users. Each node can be expanded or collapsed revealing more or less detail by clicking on the plus or minus sign to the left of the node. The user manipulates the tree to display the appropriate partition and prints the report for a summary of their findings. Each node contains a complete visual summary of the segments in that power spectrum or by contour plots of delta, theta, alpha, and beta activities. Hyper-links from the cluster nodes to the applicant's Insight EEG review software offers immediate access to pages of interest.