Multichannel Biomagnetic Artifact Removal via Signal Space Separation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multichannel biomagnetic measurement technologies face challenges in accurately removing artifacts and noise from individual sensors, which degrade the quality of measurements in MEG, EEG, and MRI due to sensitivity to calibration errors and the misinterpretation of sensor noise as neural activity.
Innovation Solution
A method that generates an n-dimensional subspace using a physical or statistical model to identify and remove artifacts by adding specific basis vectors representing individual channel signals, allowing for the decomposition and subtraction of noise components, thereby improving signal quality by isolating sensor-specific noise and interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If the SSS method is used to separate magnetic fields from different sources, then environmental interference can be removed, but sensor artifacts and calibration errors remain in the measurement signal
Solution Approach 1:
The invention segments the signal space into multiple independent subspaces: a first subspace for environmental interference fields and a second subspace for sensor artifacts. By dividing the signal decomposition into separate components, the method can independently remove each type of interference without affecting the other, thus resolving the contradiction between removing environmental interference and eliminating sensor artifacts
Solution Approach 2:
The invention extracts and removes sensor artifact components from the measurement signal by identifying them as specific subspace components. The method separates the measurement signal into multiple subspaces and extracts the artifact subspace for removal, thereby eliminating sensor artifacts while preserving the useful neural signal and environmental interference removal capabilities
2Measurement precision
If sensors are placed close to the measured object to reduce noise, then signal-to-noise ratio improves, but sensitivity to calibration errors increases
Solution Approach 1:
The invention introduces an intermediary signal processing layer that acts as a mediator between the raw sensor measurements and the final neural signal extraction. By projecting the measurement signal onto multiple subspaces and selectively removing artifact components, the intermediary processing reduces the impact of calibration errors on the final result, thus resolving the contradiction between improved signal-to-noise ratio and increased calibration error sensitivity
3Reliability
If multiple basis vectors are added to model sensor-specific signals, then artifact removal improves, but computational complexity increases
Solution Approach 1:
The invention applies partial action by adding only a limited number of basis vectors (m vectors) to the original SSS basis set, where m is typically small (e.g., 1-10). This partial extension provides sufficient artifact removal capability without requiring complete modeling of all possible sensor artifacts, thus resolving the contradiction between improved artifact removal accuracy and increased computational complexity
Data Source
AI summary
The present invention introduces a method, device and a computer program for removing artifacts present in individual channels of a multichannel measurement device. At first, a basis is generated defining an n-dimensional subspace of the N-dimensional signal space, where n is smaller than N, where using in the definition of the n-dimensional basis a physical model of a Signal Space Separation method, or a statistical model based on the statistics of recorded N-dimensional signals. Thereafter, a combined (n+m)-dimensional basis is formed by adding m signal vectors to the n-dimensional basis, each of these m signal vectors representing a signal present only in a single channel of the N-channel device. After this the recorded N-dimensional signal vector is decomposed into n+m components in the combined basis, and finally, components corresponding to the m added vectors in the combined basis are subtracted from the recorded N-dimensional signal vector.