EEG source analysis

(neurology::diagnostics)

We focus on solving the inverse problem in electroencephalography (EEG) source analysis. Surrogate-based solution techniques are employed for accelerating the solution of the inverse problem [1]. Moreover, we developed techniques for reducing the propagation of the uncertain conductivity values towards the inverse problem. For the first time in EEG source analysis (inverse problems) we were able to reduce the impact of the uncertain conductivity values (uncertainly known parameter values) upon the recovered dipole positions (inverse problem solutions). Hence, increasing the spatial resolution of EEG source analysis. The figure below shows the reduction in dipole position errors when using the subspace electrode selection methodology proposed in [2]. Further research aims at reducing the propagation errors of the uncertainty in a stochastic framework.

Dipole position errors (mm) when using the traditional methodology (left) and when using the subspace electrode selection methodology (right). The assumed conductivity ratio (uncertainty) used in the numerical procedure is different from the actual conductivity ratio (in the patient).
 
[1] G. Crevecoeur, H. Hallez, P. Van Hese, Y. D'Asseler, L. Dupré, R. Van de Walle, "A hybrid algorithm for solving the EEG inverse problem from spatio-temporal EEG data," Medical & Biological Engineering &  Computing, vol. 46, pp. 767-777, 2008.
[2] B. Yitembe, G. Crevecoeur, R. Van Keer, L. Dupré, "Reduced Conductivity Dependence Method for Increase of Dipole Localization Accuracy in the EEG Inverse Problem," IEEE Trans. Biomed. Eng., vol. 58, pp. 1430-1440, 2011.