Independent components of magnetoencephalography: localization

Independent component analysis (ICA) is a class of decomposition methods that separate sources from mixtures of signals. In this chapter, we used second order blind identification (SOBI), one of the ICA method, to demonstrate its advantages in identifying magnetic signals associated with neural information processing. Using 122-channel MEG data collected during both simple sensory activation and complex cognitive tasks, we explored SOBI’s ability to help isolate and localize underlying neuronal sources, particularly under relatively poor signal-to-noise conditions. For these identified and localized neuronal sources, we developed a simple threshold-crossing method, with which single-trial response onset times could be measured with a detection rate as high as 96%. These results demonstrated that, with the aid of ICA, it is possible to non-invasively measure human single trial response onset times with millisecond resolution for specific neuronal populations from multiple sensory modalities. This capability makes it possible to study a wide range of perceptual and memory functions that critically depend on the timing of discrete neuronal events. The goal of this chapter is to introduce the basic concept of independent component analysis (ICA), a class of algorithms that decompose a multidimensional tim