Data cleaning may be a greater challenge, as methods such as independent component analysis (ICA) for removing eye movement and other artifacts from the EEG signal may not be applicable for the sparsest of montages. For the slightly less sparse montages, separating brain from nonbrain source signals in the recorded traces may be difficult. It often involves a comparison of loading patterns of independent components onto nearby electrodes: If source activity is observed on a single EEG electrode, it cannot come from a brain source however close to the skull and dura. Brain sources project their electrical activity to an area of the scalp which generally contains several electrodes in a recording array of 30 channels or more, due to the high relative impedance of the skull compared to the brain and scalp tissues. The denser the electrode layout, the better it can be evaluated whether a source signal stems from a brain source or outside.
