SBIR-STTR Award

The spontaneous magnetoencephalogram (MEG) signals that are unrelated to a stimulus OR a task, represent a fundamental barrier to continuing brain research. This background activity constitutes a form of interference, OR "brain noise", that masks the desired weaker MEG signals evoked by the stimulus. Present techniques for enhancing the evoked response of the MEG such as signal averaging, are not applicable to single trail MEG measurements. Such measurements are important to our understanding of human performance and cognitive processing because the brain's response may be modulated by factors such as attention and work load which can change dynamically. We propose developing a new technique, termed lead field synthesis (LFS), which incorporates all the MEG signals measured from an array of sensors into a single measurement. This "virtual sensor" estimates how much of the observed MEG signal is attributable to a particular location within the brain. LFS combines aspects of spatial signal averaging, noise reduction, and inverse solution. We will develop the LFS technique for improving SNR in MEG measurements. This project represents a vital step toward the goal of low-noise imaging of brain electrical activity.

Magnetoencephalography (MEG) has been used in the study of human cognitive processing and measurement of task-related brain activity. The value of MEG to such studies lies in its ability to localize sources of activity within the brain. In practice, however, the unrelated spontaneous brain activity, or brain noise, masks the desired response so the measurement is repeated many times and the responses averaged to achieve an acceptable signal-to-noise ratio. Repetition is often impractical because of subject fatigue or modulation of attention. The problem addressed by this project is to devise signal processing techniques to reduce the interference of brain noise so the brain's response can be assessed from one or few trials. Our approach to this problem, Lead Field Synthesis (LFS), exploits the spatial redundancy afforded by multiple-sensor MEG measurements. LFS transforms the simultaneous signals from all sensors into a single virtual sensor signal having spatially selective properties. In effect, LFS focuses the sensor's response to the region of interest, and attenuates the brain noise due to sources elsewhere in the brain. The feasibility of this technique was demonstrated during Phase I. During Phase II we will develop this FLS technique and test its capabilities.