Back to Inverse Problems

Inverse EEG Applications

Figure 1

In the Inverse EEG Application, the goal is to localize an equivalent dipole source in a finite element model of a patient's head. Given EEG measurements from electrodes placed at known locations on a patient's scalp, and a discretized tetrahedral volume mesh with each labeled with the appropriate conductivity of the underlying anatomy, we solve a source localization inverse problem to identify the position and moment of the dipole within the domain which best reproduces the EEG measurements. We approach this application as an optimization problem, and have constructed the BioPSE network shown in Figure 1 to solve it. The DipoleSearch module near the top left iteratively chooses new dipole location until the algorithm converges on the optimal location.

For this example, we used a 400,000 element tetrahedral finite element model, with four different material types through the cranial domain. Using our precomputed leadfield matrix, we were able to localize the optimal dipole source in less than a minute using one processor of an sgi Origin 2000.

Figure 2

Figure 3

In Figures 2 and 3 we show the ViewWindow from the network, as the algorithm is converging. The electrodes around the outside of the head are depicted as disks with penetrating spheres. For each electrode, the measured voltage is pseudo-colormapped onto the disk surface, and the voltage due to the latest test dipole is shown pseudo-colormapped to the spheres. As the test dipoles (shown as gray arrows) iteratively approach the true dipole source (shown as a red arrow), the disks and spheres converge to have the same color.

• Publications -- Inverse EEG Application
• Image Gallery -- Inverse EEG Application

Principal Researchers:

• Dr. David Weinstein
• Rob VanUitert