What is ICA?

Independent component analysis (ICA) is a powerful linear decomposition technique aiming to uncover the sources of otherwise mixed signals. Imagine sound recordings of a cocktail party; it's often hard to follow a conversation while others are talking at the same time. Amazingly, processing such recordings with ICA allows to tease apart the different conversations - but only under certain circumstances! For instance, the recording should consist of more channels (microphones) than sound sources (people) and the sources should not move during the recording. The recording should also be taken over a reasonable interval....

My research involves the application of ICA to multi-channel EEG data. ICA is a powerful (if not the most powerful) technique for the removal of artifacts from EEG recordings. Moreover, it also helps to disentangle signals generated by different brain areas from each other. However, ICA is not a neural source localization technique, although it usually makes the identification of possible neural EEG sources easier. Like many other technologies, it requires some experience, and it's certainly most helpful to have a reasonable understanding about the physiology of the EEG signal.