Inclusion of temporal constraints in the EEG inverse problem: A comparative study

Electroencephalographic (EEG) recordings contain dynamic data with strong temporal information, therefore it is important the inclusion of such information in the inverse problem solution to obtain an improvement in estimation of the brain activity, reducing noise effects and guaranteeing a smooth temporal evolution. This work presents and compare two ways to include the temporal information in the inverse problem solution: On the one hand, a temporal projection over the main dynamics of the data. On the other hand, an explicit model to describe the temporal evolution of the brain activity. The performance of the proposed method is evaluated using simulated EEG data under several SNRs in terms of spatial accuracy, temporal accuracy, the mean squared error and computational cost. Obtained results show that the inclusion of temporal information using a temporal projection is more robust to noise and its computational cost is significantly lower than the explicit temporal model.