hide
Free keywords:
Extracerebral compartments; Transcranial Magnetic Stimulation; Transcranial Electrical Stimulation; Electroconvulsive Therapy
Abstract:
Extracerebral brain compartments comprise partitioning skull into outer table (cortical bone), diploë (cancellous or trabecular bone), inner table (cortical bone), three meningeal layers (dura, arachnoid, and pia mater tightly enclosing the brain) as well as partitioning scalp into several tissue types such as skin itself, muscle, and fat. At present, their accurate segmentation is difficult. However, meaningful results could be obtained by expanding the existing subject specific automated brain segmentation of major compartments using anatomical rules known from many prior extensive studies.
Extracerebral tissue types present additional conductivity interfaces which indeed distort transcranial stimulation fields, most notably the electric field. What is the order of this distortion for different modalities and targets? Although an approach exists which replaces this effect by modified conductivities of simpler models, this is true only for specific subjects/locations/montages where an approximate adjustment can be made via accurate numerical modeling of a particular problem.
Modeling tightly spaced thin layers of tissues is computationally expensive when using the finite element method (FEM) due to the requirement of an adequate volumetric mesh. To overcome this issue, we use the boundary element fast multipole method (BEM-FMM), which only requires the discretization of the surfaces. Furthermore, we extend the existing BEM-FMM algorithm by an automated adaptive mesh refinement mechanism. With this extension, we are able to determine accurate and self-converged results for the electric field with reasonable computational cost.
Our extensive modeling results reveal that the extracerebral brain compartments are less important for TMS but are quite important for TES and ECT dose predictions. There and for deeper targets, the electric field deviation could be on the order of 100%. Our method could be applicable to any head geometry with or without lesions, and also to tissues in the form of thin separate islands.
