EMBRIOSS: ELECTROMAGNETIC BRAIN IMAGING BY OPTIMIZATION IN SPECTRAL SPACE

We propose a new method called electromagnetic brain imaging by optimization in spectral space (EMBRIOSS) for electromagnetic source imaging of the human brain. The method incorporates the physiological knowledge that electrical activities in the brain are spatially coherent and often sparse. For spatial coherency, we confine the solution to a subspace spanned by the low-frequency eigenvectors of a Laplacian of the cortical surface mesh. For sparseness, we apply a p-norm regularization with p < 2. The resulting nonlinear regularization problem is solved efficiently using half-quadratic programming. Through realistic simulations, we have compared our method with existing approaches. The results show that our method performs better