Dipole estimation errors due to not incorporating anisotropic conductivities in realistic head models for EEG source analysis

• Published in: Physics in medicine & biology 2009-10, Vol.54 (20), p.6079-6093
• Main Authors: Hallez, Hans, Staelens, Steven, Lemahieu, Ignace
• Format: Article
• Language: English
• Subjects: Algorithms; Anisotropy; Brain - pathology; Computer Simulation; Electrodes; Electroencephalography - methods; Head - physiology; Humans; Models, Neurological; Nervous System Diseases - diagnosis; Nervous System Diseases - pathology; Reproducibility of Results; Signal Processing, Computer-Assisted; Skull - anatomy & histology; Skull - pathology
• ISSN: 0031-9155; 1361-6560
• DOI: 10.1088/0031-9155/54/20/004

EEG source analysis is a valuable tool for brain functionality research and for diagnosing neurological disorders, such as epilepsy. It requires a geometrical representation of the human head or a head model, which is often modeled as an isotropic conductor. However, it is known that some brain tissues, such as the skull or white matter, have an anisotropic conductivity. Many studies reported that the anisotropic conductivities have an influence on the calculated electrode potentials. However, few studies have assessed the influence of anisotropic conductivities on the dipole estimations. In this study, we want to determine the dipole estimation errors due to not taking into account the anisotropic conductivities of the skull and/or brain tissues. Therefore, head models are constructed with the same geometry, but with an anisotropically conducting skull and/or brain tissue compartment. These head models are used in simulation studies where the dipole location and orientation error is calculated due to neglecting anisotropic conductivities of the skull and brain tissue. Results show that not taking into account the anisotropic conductivities of the skull yields a dipole location error between 2 and 25 mm, with an average of 10 mm. When the anisotropic conductivities of the brain tissues are neglected, the dipole location error ranges between 0 and 5 mm. In this case, the average dipole location error was 2.3 mm. In all simulations, the dipole orientation error was smaller than 10 degrees . We can conclude that the anisotropic conductivities of the skull have to be incorporated to improve the accuracy of EEG source analysis. The results of the simulation, as presented here, also suggest that incorporation of the anisotropic conductivities of brain tissues is not necessary. However, more studies are needed to confirm these suggestions.