fNIRS Motion Artifact Removal via Channel Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for removing motion artifacts from functional near-infrared spectroscopy (fNIRS) signals, such as wavelet description length detrending, fail to consider spatial features of multi-channel signals, leading to attenuation or distortion of hemodynamic response signals and difficulties in distinguishing task-related components from motion artifacts, especially in single task-based environments.
Innovation Solution
A method involving the calculation of correlation coefficient differences between channels to identify and isolate motion artifacts, followed by wavelet transformation and neural network-based weight adjustment to remove artifacts from affected channels, utilizing an arrangement of optodes on the scalp to form channels and applying weights to neighboring channels to mitigate motion artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If wavelet description length detrending technique is used to remove motion artifacts from fNIRS signals, then motion artifacts are reduced, but hemodynamic response signal components are attenuated or distorted
Solution Approach 1:
The patent segments the fNIRS signal processing by separating motion artifact removal from hemodynamic response analysis. It applies wavelet transformation to decompose signals into different frequency bands, then selectively processes motion artifact components while preserving hemodynamic response components through channel-specific correlation analysis and targeted detrending operations.
Solution Approach 2:
The patent applies local quality by treating each channel individually with unique motion artifact characteristics. It calculates channel-specific correlation coefficients between neighboring channels to identify motion artifacts locally, then applies detrending operations only to affected channels or specific time segments, preserving the integrity of non-affected signal portions.
2Object-affected harmful factors
If wavelet description length detrending technique is applied to single task-based experiments, then motion artifacts are removed, but task-related hemodynamic components are misidentified and removed
Solution Approach 1:
The patent implements feedback mechanisms by continuously monitoring channel correlation coefficients during signal processing. It uses real-time correlation analysis between neighboring channels to detect motion artifacts, then adjusts detrending operations accordingly, providing feedback loops that prevent the removal of task-related hemodynamic components while eliminating motion artifacts.
Solution Approach 2:
The patent changes processing parameters dynamically based on signal characteristics. It adjusts wavelet decomposition levels, correlation threshold values, and detrending intensity according to the specific experimental conditions and signal quality, allowing flexible adaptation to single task-based experiments while preserving task-related components.
Data Source
AI summary
Disclosed is a method of removing motion artifacts from functional near-infrared spectroscopy (fNIRS) signals. The method includes: disposing N optodes at a plurality of locations on a scalp, and forming a plurality of channels between the N optodes; calculating a correlation coefficient difference index of a neighbor channel around each receiver optode, and detecting a receiver optode in which a motion artifact has occurred based on the correlation coefficient difference index; and removing motion artifacts based on the detected receiver optode and an arrangement structure of the N optodes.