LORETA can be done in a few ways… as a source analysis for the entire EEG or segment being analyzed… or after a decomposition into the ICA components… The later will provide a better estimate of generators of the
individual components which are all blended for the overall LORETA, distorting the sources with other sources.
I would suggest looking at the ICA of the cleaned up EEG, and then finding the component that matches the topography and spectral features, and time frame as the event you want to image… and then doing the LORETA of only that component…
The solution of the mixed components in the raw file will blend many components, and if the one you are looking for is weaker, it will be swamped with the other sources… but this does not mean it is insignificant
functionally…. just harder to image from the raw blended component mix.