Multi-Scale Modeling for Analysis and Design of Transcranial Electric and Magnetic Brain Stimulation
Transcranial electric stimulation (tES) and magnetic stimulation (TMS) can noninvasively modulate brain activity in humans, offering broad research and therapeutic applications. However, improving the efficacy and selectivity of these techniques is challenging without a mechanistic understanding of how the stimulation parameters determine the neural response and how these parameters can be manipulated to activate specific neural circuits. This dissertation presents multiscale computational models that predict the neural response to TMS and tES at the single-cell and population levels for analysis and rational design of transcranial brain stimulation.We adapted biophysically-realistic models of cortical neurons from the Blue Brain network to the properties of mature rat and human neurons and characterized the direct response to extracellular stimulation with both subthreshold and suprathreshold electric field (E-field) stimulation modalities. These models included 3D reconstructed axonal and dendritic arbors as well as multiple excitatory and inhibitory cell types with validated electrophysiological behavior. Axon terminals were the lowest threshold elements for stimulation, and their dependence on threshold and polarization was determined by cell-type specific morphological features, such as myelination, diameter, and branching. However, we found for TMS pulse waveforms specifically, activation thresholds were higher than expected from in vivo applications. We improved the fidelity of the axon models further using a feature-based optimization algorithm, but these modifications did not produce models with significantly lower thresholds, suggesting other factors may allow for suprathreshold activation at the E-field intensities induced experimentally. The neuron models were then embedded in anatomically-realistic volume conductor head models of the E-field in humans derived from magnetic resonance imaging (MRI) data to simulate the direct neural response to TMS and tDCS. The models reproduced relative trends in motor thresholds as well as the experimentally-measured strength–duration time constant. TMS activated with lowest intensity intracortical axon terminals in the superficial gyral crown and lip region, proportional to the E-field magnitude. Thresholds were lowest for the L5 pyramidal cells (PCs), with activation of the L2/3 PCs and large basket cells at most intensities. Reversing the pulse direction revealed waveform-dependent spatial shifts in the activated neural population that may explain experimentally observed differences in the latencies and thresholds of muscle responses to TMS of motor cortex. We also quantified the subthreshold polarization generated by conventional tDCS with large rectangular pad electrodes and 4×1 high definition (HD) tDCS electrodes targeting the motor hand knob. Axonal and dendritic terminal polarization was higher than somatic polarization in all cell types, and polarization trends between cell types varied by subcellular compartment. While the HD tDCS montage produced a significantly more focal E-field within the brain, both montages generated broad regions of depolarization and hyperpolarization beneath the electrodes. These simulations demonstrated the importance of coupling the E-field to neuron models incorporating non-linear membrane dynamics and realistic morphologies for predicting the neural response to TMS and tES. Indeed, extrapolating the neural response (polarization or threshold) from the uniform E-field or macroscopic E-field components often led to erroneous predictions. Due to the high computational cost of these biophysically-realistic models, we also developed rapid estimators of the neural response using a 3D convolutional neural network. This approach allowed for reproducing the threshold distributions of the realistic model neurons with several orders of magnitude shorter run times than using the E-field distribution alone. In sum, this work provides both computational tools and mechanistic insights to improve the use and development of transcranial magnetic and electrical stimulation technologies.
Finite element method
Transcranial electric stimulation
Transcranial magnetic stimulation
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations
Works are deposited here by their authors, and represent their research and opinions, not that of Duke University. Some materials and descriptions may include offensive content. More info