EEG Noise Interval Detection via Frequency Power Baseline
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting and rejecting noise in biosignals, particularly EEG analysis, face challenges in accurately distinguishing noise related to motion and drowsiness, which can lead to errors in brainwave analysis and biomarker determination due to the diversity of noise types and causes.
Innovation Solution
An automatic noise signal interval detection method and device that converts time-series data into a frequency domain, calculates power for each epoch, generates a power graph, sets a baseline, and determines noise signal intervals exceeding a threshold, allowing for the rejection of noise intervals and improving analysis reliability and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual noise detection and rejection methods are used in EEG analysis, then expertise and manual intervention are required to identify noise patterns, but this leads to reduced productivity and increased time consumption for analyzing biosignals
Solution Approach 1:
The system performs automatic noise detection and rejection without requiring manual expert intervention. The noise detection unit automatically identifies noise patterns in EEG signals by analyzing frequency domain characteristics, and the signal correction unit automatically corrects or rejects contaminated segments, enabling the system to serve itself in the noise rejection process
Solution Approach 2:
The patent replaces manual mechanical analysis methods with automated computational algorithms. Instead of experts manually examining time-series EEG data to identify noise patterns, the system uses frequency domain transformation and automated pattern recognition algorithms to detect and reject noise, substituting human expertise with computational processing
2Productivity
If automated noise rejection algorithms are implemented, then productivity and consistency are improved, but the ability to accurately distinguish diverse noise types (motion artifacts, drowsiness, muscle activity) remains insufficient
Solution Approach 1:
The system transforms the noise detection problem from the time domain to the frequency domain using Fast Fourier Transform. By analyzing EEG signals in the frequency domain, the system can identify characteristic frequency patterns of different noise types (e.g., muscle artifacts in beta band, drowsiness-related theta waves), adding a frequency dimension to the analysis that enables better discrimination of diverse noise types
Solution Approach 2:
The patent changes the analysis parameters by examining frequency domain characteristics rather than time-domain waveforms. The noise detection unit analyzes power spectral density and frequency distribution patterns to identify noise, using parameters such as frequency power ratios and spectral characteristics that vary differently across noise types, enabling more accurate automated discrimination
3Reliability
If frequency domain analysis is used to detect noise patterns, then noise detection capability is improved, but the complexity of the analysis process increases
Solution Approach 1:
The patent segments the noise rejection process into distinct functional modules: a noise detection unit that identifies noise patterns in the frequency domain, and a signal correction unit that applies appropriate corrections or rejection. This segmentation allows each module to perform its specific function with optimized algorithms, managing overall system complexity while maintaining high reliability
Data Source
AI summary
An automatic noise signal interval detection method and device are provided, and the method includes receiving input data and generating initial data in a time-frequency domain, calculating power for each epoch for each channel of the initial data, generating a power graph for a specific frequency region of each channel based on the power for each epoch, generating a baseline based on an average of each channel value on the power graph, and determining, as a noise signal interval, an interval exceeding a predetermined threshold based on the baseline on the power graph.