Abstracts

Rationale: The objective of this presented work is to predict preoperatively the extent to which direct stimulation therapy (NeuroPace^TM) can propagate through pathological white matter during responsive neurostimulation (RNS). A presurgical model was generated to calculate the volume of cortical activation (VOCA) in an RNS system candidate. This model comprises an iterative computationally-intensive process that tests available stimulation parameters for bipolar depth electrodes. Such a model includes a proposed method to calculate potential regions of hyperpolarization and depolarization at the grey-white matter interface using an activation function. Methods: An SPGR MRI sequence, interictal diffusion tensor imaging (DTI) dataset, and post-ictal DTI were acquired for our candidate. DTI sequences were obtained on a 3T Siemens^TM MRI scanner using 2mm- thick oblique axial slices. Diffusion measurements were performed in 60 non-collinear directions with a diffusion weighting factor of 900 s/mm². Six non-diffusion weights (b-values) were used with a repetition time of 10s. Ictal propagation pathways were determined using novel subtracted post-ictal diffusion tensor imaging (spiDTI) developed at our center. This technique was performed by computing fractional anisotropy (FA) in interictal and post-ictal sequences and subtracting the results. A threshold of three std dev was used to differentiate substantial changes in FA. The iterative process was developed as follows: (1) Using our planning system, depth leads were strategically positioned near the spiDTI signal, in a 3D segmented mesh generated from the MRI scan. (2) The extracellular electric potential (EP) predicted by electric stimulation modeling was calculated using the finite element method in a homogeneous isotropic medium (COMSOL Multiphysics^TM). (3) The EP model was used to estimate an activation function for the cable equation of axons by computing the second directional derivative in the direction of white matter tracts. This directionality was obtained from tensor fitting acquired from the post-ictal DTI (MATLAB^TM). (4) The magnitude of the activation function was used to determine areas of hyperpolarization and depolarization adjacent to the electrodes. (5) Generated depolarization/hyperpolarization regions of interest were used to identify influenced tracts in our patient-specific modulated circuit tractography model. Results: The application of the activation function in the presurgical model produced a VOCA defined as non-spherical overlapping regions of interest around the adjacent electrode where hyperpolarization and depolarization were determined. Conclusions: This dataset demonstrates the ability to generate a preoperative electrode planning map for predicting ‘best-implant' sites for RNS depth leads. Furthermore, the application of the activation function is an improvement to previously reported activation regions produced by the magnitude of electric fields (Rossi et.al 2010). This new model considers not only the magnitude of the field, but also its directionality effects in relation with the axon bundle orientation.