Harmony: EEG/MEG source reconstruction in the neural basis of spherical harmonics

EEG/MEG source localization based on minimum-norm estimate is hindered by the two problems: the localization is non-unique (since the number of sources is much larger than the number of signals) and unstable (the solution is highly sensitive to sensor noise). A new source localization algorithm is presented, which addresses both problems in a simple manner. By using the neural basis of spherical harmonics (or spherical splines) instead of single dipoles the dimensionality of the inverse solution is greatly reduced without sacrificing the quality of the data fit. The smoothness of the resulting solution reduces the surface bias and spurious scattered sources, which plague many existing minimum-norm techniques, and allows to efficiently suppress the sensor noise. The algorithm, termed Harmony, showed excellent performance when applied to real data and showed better localization precision and solution coherence than other tested algorithms when applied to realistically simulated data.