Evaluation of a finite-element reciprocity method for epileptic EEG source localization: Accuracy, computational complexity and noise robustness

Abstract

Purpose: The aim of this paper is to evaluate the performance of an EEG source localization method that combines a finite element method (FEM) and the reciprocity theorem. Methods: The reciprocity method is applied to solve the forward problem in a four-layer spherical head model for a large number of test dipoles. To benchmark the proposed method, the results are compared with an analytical solution and two state-of-the-art methods from the literature. Moreover, the dipole localization error resulting from utilizing the method in the inverse procedure for a realistic head model is investigated with respect to EEG signal noise and electrode misplacement. Results: The results show approximately 3% relative error between numerically calculated potentials done by the reciprocity theorem and the analytical solutions. When adding EEG noise with SNR between 5 and 10, the mean localization error is approximately 4.3 mm. For the case with 10 mm electrode misplacement the locali zation error is 4.8 mm. The reciprocity EEG source localization speeds up the solution of the inverse problem with more than three orders of magnitude compared to the state-of-the-art methods. Conclusions: The reciprocity method has high accuracy for modeling the dipole in EEG source localization, is robust with respect to noise, and faster than alternative methods

A finite-element reciprocity solution for EEG forward modeling with realistic individual head models

“…The direct method, in contrast, requires thousands of iterations to create a leadfield from a dense array of sources. A number of previous studies have demonstrated the use of reciprocity for FEM solutions [22,2,3,7]. This article draws heavily on the work of Weinstein et al for its basis [22].…”

“…In this paper we focus on the forward problem, which has not been given a great deal of attention in neuroimaging. Despite many published studies demonstrating high-quality head models and forward modeling approaches [2,3,4,5,6,7,8], most functional neuroimaging studies still rely on relatively poor quality electromagnetic head models when performing source localization.…”

High-Resolution EEG Source Reconstruction with Boundary Element Fast Multipole Method Using Reciprocity Principle and TES Forward Model Matrix

“…In EEG/MEG analyses, the reciprocity principle has been previously used for BEM [27], [28], [32], FDM [29], [28], [32], and FEM [31], [32], [33], [34] methods, but its applications have generally been limited. This is perhaps because, for identical head models, both conventional (dipole-based) and reciprocal (electrode-based) approaches are very similar in the final result for EEG applications, as they both change the forward problem from a source point of view to a sensor point of view [31].…”

“…The goal of this task is to estimate an average noiseless EEG source localization floor for a deeplylocated tangential cortical dipole cluster, with the response resembling that of the N20/P20 peak. Some previous studies reported very high ideal source reconstruction accuracy, such as twice the size of the discretization element [33]. However, these studies were either restricted to spherical head models [28], [29], [30], [33] and/or to one subject [29], [33].…”

“…Some previous studies reported very high ideal source reconstruction accuracy, such as twice the size of the discretization element [33]. However, these studies were either restricted to spherical head models [28], [29], [30], [33] and/or to one subject [29], [33]. Sometimes, the exact source placement was not entirely clear [33].…”