Minimum L1 norm MEG reconstruction minimising signal deviation using a reduced lead field

The MEG sources are localized with a weighted minimum L1 norm using a size reduced lead field and minimizing the difference between measured and calculated signals. A singular value decomposition (SVD) is used for reducing the lead field matrix. The size reduced lead field is well suited for reconstructing sparse sources. With a fixed lead field matrix, the difference between the measured data and calculated signals depends on the weights of the L1 norm. The weights are successfully chosen such that the signal difference is minimized. Examples illustrate the approach.