Neural networks for reconstruction of focal events from bioelectric/biomagnetic potentials
Martin F. Schlang, Michael Haft, K. Abraham‐Fuchs · 1994
When analysing bioelectric or biomagnetic fields source reconstruction plays an important role. We show how the conventional reconstruction of dipole sources can be speeded up by neural networks. In this paper several different types of neural networks and their performance are compared for this task. We show that a radial basis function network with partitioning to one yields the best recognition results especially for potentials which are superimposed with noise.>