Kinetic images of neuronal activity of the human brain based on the spatio-temporal MNLS inverse: a theoretical study.

Prior work proved that it is possible to find a unique solution to the problem of defining the configuration of electric current underlying observed extracranial magnetic fields, if sufficient priori knowledge of the source configuration is available. This minimum-norm least-squares (MNLS) inverse solution for a magnetic source image (MSI) is extended here to include temporal as well as spatial parameters of the underlying current pattern. This capitalizes on the temporal resolution of magnetoencephalography (MEG), which is on the order of milliseconds. Other forms of functional brain imaging are far less sensitive to the rate of change of states of the brain. Influences on the quality of the resulting MSI by measurement noise and errors in determining the image surface are characterized. A new technique for reducing noise in the inverse problem is developed by taking into consideration the spatial and time-dependence of the noise detected by the sensors. This new approach to regularization reduces the contribution from noisy measurements to the inverse calculation, and therefore improves the stability of the inverse.