Multi-Core Beamformer for Correlated MEG Source Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional magnetoencephalography (MEG) source-modeling methods struggle to accurately localize highly-correlated neuronal networks from noisy data, as they assume uncorrelated source time-courses, leading to suppression of correlated sources and difficulty in identifying entire neural activity pathways.
Innovation Solution
The Multi-Core Beamformer (MCBF) technique reconstructs source power covariance matrices to determine individual time-courses and correlations, using an enhanced dual-core beamformer (eDCBF) with a single computation of a weight matrix to handle correlated and uncorrelated sources simultaneously, reducing computational time and improving source localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional beamformer methodology is used to filter MEG sensor signals, then computational cost is low and the method works well for low SNR data, but source-power estimates from highly correlated source-grid dipoles are suppressed
Solution Approach 1:
The patent segments the beamformer computation into multiple cores, where each core processes a subset of sensor signals independently to compute partial source-power estimates. These partial estimates are then combined to reconstruct the full source activity, including highly correlated sources that would be suppressed in conventional single-core beamforming.
Solution Approach 2:
The patent merges results from multiple independent beamformer computations by combining source-power estimates from different cores. This combining process reconstructs the complete source activity pattern, including correlated neural networks, while maintaining the computational efficiency of individual beamformer operations.
2Device complexity
If conventional beamformer assumes uncorrelated source time-courses, then computation is simplified, but correlated neural networks cannot be accurately reconstructed
Solution Approach 1:
The patent segments the computational process into multiple independent beamformer cores, each making the uncorrelated source assumption locally. This segmentation allows each core to compute efficiently while the overall system, through combination of results, accurately reconstructs correlated sources without requiring complex inter-core communication during computation.
Solution Approach 2:
The patent introduces an intermediary combining step that aggregates results from multiple simple beamformer computations. This intermediary process reconciles the simplified uncorrelated-source assumptions made in each core with the need to accurately represent correlated neural networks in the final reconstruction.
3Measurement precision
If multiple beamformers are used to reconstruct correlated sources, then source reconstruction accuracy improves, but computational time increases significantly
Solution Approach 1:
The patent segments the multiple beamformer computations into parallel independent cores that can be executed simultaneously. This segmentation transforms sequential computational time into parallel processing time, maintaining high reconstruction accuracy while reducing total computational time through concurrent execution of multiple beamformer instances.
Solution Approach 2:
The patent implements periodic independent beamformer computations across multiple cores, where each core periodically processes its assigned sensor subset and contributes to the final reconstruction. This periodic parallel action achieves accurate correlated source reconstruction without the continuous computational burden of a single complex beamformer.
Data Source
AI summary
Techniques and systems are disclosed for implementing multi-core beamforming algorithms. In one aspect, a method of implementing a beamformer technique includes using a spatial filter that contains lead-fields of two simultaneous dipole sources rather than a linear combination of the two to directly compute and obtain optimal source orientations and weights between two highly-correlated sources.