Abstract Across all electrical stimulation (neuromodulation) domains, conventional analysis of cell polarization involves two discrete steps: i) prediction of macroscopic electric field, ignoring presence of cells and; ii) prediction of cell polarization from tissue electric fields. The first step assumes that electric current flow is not distorted by the dense tortuous network of cell structures. The deficiencies of this assumption have long been recognized, but – except for trivial geometries – ignored, because it presented intractable computation hurdles. We leverage: i) recent electron microscopic images of the brain that have made it possible to reconstruct microscopic brain networks over relatively large volumes and; ii) a charge-based formulation of boundary element fast multipole method (BEM-FMM) to produce the first multiscale stimulations of realistic neuronal polarization by electrical stimulation that consider current flow distortions by a microstructure. The dataset under study is a 250×140×90 μm section of the L2/L3 mouse visual cortex with 396 tightly spaced neurite cells and 34 microcapillaries. We quantify how brain microstructure significantly distorts the primary macroscopic electric field. Although being very local, such distortions constructively accumulate along the neuronal arbor and reduce neuronal activating thresholds by 0.55-0.85-fold as compared to conventional theory. Data availability statement Post-processed cell CAD models (383), microcapillary CAD models (34), post-processed neuron morphologies (267), extracellular field and potential distributions at different polarizations (267×3), *.ses projects files for biophysical modeling with Neuron software (267×2), and computed neuron activating thresholds at different conditions (267×8) are made available online through BossDB, a volumetric open-source database for 3D and 4D neuroscience data. Significance statement This study introduces a novel method for modeling perturbations of impressed electric fields within a microscopically realistic brain volume, including densely populated neuronal cells and blood microcapillaries. It addresses a limitation present across decades of macroscopic-level electromagnetic models for electrical stimulation. For the investigated brain volume, our model predicted a neural activation threshold reduction factor of 0.85–0.55 when compared to the macroscopic approach. The present study begins to bridge a long-recognized gap in our analysis of bioelectricity and provides a framework to evaluate (and compensate) for the adequacy of macroscopic models in brain stimulation and electrophysiology.

Electromagnetic Modeling within a Microscopically Realistic Brain – Implications for Brain Stimulation