FIGURE 1.

Ultra high field 7 Tesla MR images for patient-specific STN parcellation, lead location, and computational modeling. (A) STN parcellation results. (B) Representation of white matter tracts between the STN and the cortex. (C,D) Lead locations with respect to the STN parcellation. (E) Patient-specific computational model of bilateral STN-DBS settings before and after right lead revision. The anatomical portion of the model was constructed from segmentation of high field imaging data (7T) with post-operative CT scans for lead localization. The STN volumes were then populated with biophysical multi-compartment neuron models that were perturbed with clinical DBS waveforms whose amplitudes were calculated from simulations of the tissue voltages induced through an anisotropic and inhomogeneous finite element model (FEM, COMSOL Multiphysics 5.4) of the electrode-tissue interfaces for this patient. The FEM was parameterized using diffusion-weighted imaging data from the patient. These models provided a quantitative estimate of the percentage of each neuronal pathway directly modulated by a clinical stimulation setting, with maximum possible activation of 100% for each pathway. Across all lead implants, the patient-specific models showed that stimulation of the motor STN was important to treat parkinsonian motor signs, while stronger activation of the associative and limbic territories resulted in the acute effects on mood.