Simultaneous estimation of brain structure and function with MEG/EEG data

The traditional reconstruction of the sources of brain activity from MEG/EEG requires to previously acquire a structural image estimated from a magnetic resonance imaging scan. Then, the neural activity is mapped into this anatomical model by solving an inverse problem. However, obtaining structural images is expensive and time consuming. A widely used alternative is to use standard brains from public databases, but this procedure is inaccurate due to the large differences in sulci shape and distribution among subjects. Magneto- and electro-encephalography (M/EEG) data are desirable as they are passive and non-invasive with high temporal resolution. We have developed a methodology to make estimates of brain structure based on MEG/EEG data. We do this by estimating the most plausible brain structure fitted to a given dataset. Although there are many possible cortical structures that could give rise to any single segment of MEG/EEG data, there is only one anatomy that can support all of the measured data. With this methodology, we can obtain better MEG/EEG brain images than with a standard brain, without the need of acquiring a subject specific structural image. Additionally, we can simultaneously provide a more accurate subject specific anatomical model than a standard one.