Particle Swarm Optimization Applied to EEG Source Localization of Somatosensory Evoked Potentials

Abstract

One of the most important steps in presurgical diagnosis of medically intractable epilepsy is to find the precise location of the epileptogenic foci. Electroencephalography (EEG) is a noninvasive tool commonly used at epilepsy surgery centers for presurgical diagnosis. In this paper, a modified particle swarm optimization (MPSO) method is used to solve the EEG source localization problem. The method is applied to noninvasive EEG recording of somatosensory evoked potentials (SEPs) for a healthy subject. A 1 mm hexahedra finite element volume conductor model of the subject's head was generated using T1-weighted magnetic resonance imaging data. Special consideration was made to accurately model the skull and cerebrospinal fluid. An exhaustive search pattern and the MPSO method were then applied to the peak of the averaged SEP data and both identified the same region of the somatosensory cortex as the location of the SEP source. A clinical expert independently identified the expected source location, further corroborating the source analysis methods. The MPSO converged to the global minima with significantly lower computational complexity compared to the exhaustive search method that required almost 3700 times more evaluations.

Unsupervised Segmentation of Head Tissues from Multi-modal MR Images for EEG Source Localization

“…For data sets 2, 3, and 4, a radio-oncologist, with more than 17 years experience, manually segmented each subject data into five tissue classes: WM, GM, CSF, skull and skin. This represented a tradeoff between the time needed to manually segment the images and delineating the essential classes for EEG source localization (these five classes are commonly used for EEG source localization [4,5,7]). The radio-oncologist included fat, muscle, and skin in the "skin" class.…”

“…Synthetic EEG was generated using the 2D ground truth by placing a source in the GM tissue and calculating EEG signals from 30 electrodes placed equidistantly around the head, based on the 10/10 system [1,5,7]. Implementation Details and Optimal Parameters Settings for the Segmentation Algorithms MATLAB, together with FSL, was used to implement all of the algorithms.…”

“…The localization procedure is that described in [1,5,7]. In summary, given a head conductivity model and the EEG data, the procedure involves the solution of the following two problems: (i) the forward problem that deals with finding the scalp potentials for the given current sources and (ii) the inverse problem that deals with estimating the sources to fit the given potential distributions at the scalp electrodes.…”

“…The EEG source localization performances of the proposed method HSA-BAMS and the reference method BET-FAST were evaluated quantitatively in terms of relative error (RE) of potential and localization error (LE) [1,5,7]. The relative error is defined:…”

“…It is common practice to use five tissue classes-white matter (WM), gray matter (GM), cerebrospinal fluid (CSF), skull, and skin [4,5] followed by manual correction for source localization, although improvements in localization accuracy can be achieved with more classes [6]. Completely manual segmentation [7], typically performed by a clinical expert, is the gold standard. Nevertheless, it is laborious and time consuming.…”