Fusion of Functional Brain Imaging Modalities via Linear Programming

Fusion algorithms are employed in an attempt to construct a spatio-temporal estimate of neuronal activity using data gathered from multiple functional brain imaging modalities. Here, the estimate is built by placing a dipole in each voxel of the modality with highest spatial resolution, and estimating the time course of each dipole without constraining dipole orientations. The solution space thus consists of a matrix S of dimensionality 3N×T , which actually consists of 3 N×T matrices. Each such sub-matrix corresponds to the projection of the dipole to one axis [1]: