Accelerating eigenvector computations for inverse electrocardiography applications
James Schelert, Lorraine G. Olson, M. Muff, Robert D. Throne · 2002
In previous studies on the inverse problem of electrocardiography, Generalized Eigensystem (GES) methods performed better than more conventional techniques such as singular valve decomposition or Tikhonov regularization. Here, the authors introduce Geometric Eigenvector Mapping (GEM) as a technique for accelerating the eigenvector computations which are the basis for the GES methods. The GEM method provided substantial reductions in eigenvector computation time with improvements ranging from 22% to 74%.