EEG signal feature extraction method, system, storage medium and terminal

By using multi-scale spectral structure change scoring and Hermite function decomposition, the adaptability and robustness issues of existing EEG signal feature extraction methods in sleep apnea detection are solved. This achieves efficient feature extraction and stable feature vector construction of sleep EEG, improving the accuracy and sensitivity of sleep apnea detection.

CN121997032BActive Publication Date: 2026-06-16SHANGHAI ADVANCED RES INST CHINESE ACADEMY OF SCI
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHANGHAI ADVANCED RES INST CHINESE ACADEMY OF SCI
Filing Date
2026-04-07
Publication Date
2026-06-16

AI Technical Summary

Technical Problem

Existing EEG signal feature extraction methods are unable to adapt to the dynamic changes in sleep EEG, especially during short-term emergencies such as micro-awakening, sympathetic activation, and high-frequency energy surges. They cannot accurately reflect physiological processes and lack a joint characterization mechanism of multi-scale spectral structure and emergencies, resulting in poor feature extraction robustness and difficulty in meeting the needs of automatic clinical sleep apnea screening.

Method used

By employing multi-scale spectral structure change scoring and Hermite function decomposition, and constructing multi-scale reference and test windows, candidate segmentation boundary points are detected to obtain adaptive EEG segmentation. Hermite function decomposition is then performed, and feature vectors are constructed by combining global EEG statistics to achieve adaptive segmentation and stable feature extraction of EEG signals.

Benefits of technology

It significantly improves the robustness and event sensitivity of EEG signal feature extraction, enhances the recognition accuracy and feature stability of sleep apnea-related events, solves the problem of event truncation and information dilution caused by fixed windows, enhances the detection sensitivity of short-term key events such as micro-awakening, and ensures the cross-segment consistency and generalization of features.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121997032B_ABST
    Figure CN121997032B_ABST
Patent Text Reader

Abstract

The application provides an EEG signal feature extraction method, system, storage medium and terminal, the method comprises the following steps: preprocessing the EEG signal; constructing a multiscale reference window and a test window, and calculating the multiscale spectrum structure change score of the preprocessed EEG signal; detecting the candidate segmentation boundary point of the preprocessed EEG signal based on the multiscale spectrum structure change score; obtaining the EEG adaptive segmentation based on the candidate segmentation boundary point; performing Hermite function decomposition on the EEG signal in the EEG adaptive segmentation to obtain the EEG segment-level feature; and constructing the EEG feature vector based on the EEG segment-level feature and the EEG global statistic quantity. The EEG signal feature extraction method, system, storage medium and terminal can extract the EEG signal feature with the characteristics of strong event sensitivity, high adaptive capacity and excellent stability.
Need to check novelty before this filing date? Find Prior Art