Abstracts

Rationale: A critical step toward optimizing direct neuromodulation of refractory focal-onset epileptogenic networks is to effectively interface with the epileptogenic circuit. A challenge to achieving  this goal is that the two FDA-approved direct stimulation systems (DBS, Medtronic & RNS, NeuroPace) are limited to 2 4-contact depth leads. Our objective was to predict, preoperatively, the extent to which responsive activated depth contacts can propagate through an epileptogenic brain circuit by placing virtual electrodes at the grey-white matter interface.

A classical approach to determine the volume of brain activation is to simply calculate the electric field (E-field) immediately surrounding the stimulating electrode and select a “magnitude threshold” above which cortical activation occurs. While this model gives information on possible activated tracts when placed in white matter (WM), it fails to account for stimulation effects on the axonal membrane potential. However, an understanding of the membrane biophysics is crucial for predicting activation in axon bundles adjacent to the electrode.

Our model can differentiate between regions of depolarization (DP) & hyperpolarization (HP) produced by applied stimulation by computing an activation function (AF), derived from the core-conductor model which considers three factors: 1) electric potential (EP), 2) directionality of the E-field, & 3) axon bundle orientation.

The model was generated for 8 RNS patients with refractory focal-onset epilepsy implanted at RUMC. The AF was computed for each patient and then compared with the classical E-field model.

Methods: The workflow consisting of 6 steps was completed for 8 candidates for either RNS or DBS depth lead implantation as follows:

1) Strategic virtual electrode lead placement was performed using a patient-specific 3D-model from the structural MRI; 2) computations of the EP and E-Field were completed using the finite element method; 3) the AF was calculated as the second directional derivative of the EP in the direction of axon bundles relative to the electrode; 4) regions with high E-Field DP & HP were used as seeds for creating modulated circuit tractography (MCT) maps for each model; 5) MCT maps were then used as targets for intraoperative placement of depth leads.

Results: For the 8 patients (5 RNS, 3 DBS), the AF model generated irregular volumes of activation surrounding the depth contacts due to DP & HP and compared with the spherical shape regions generated by the E-field model. The MCT activated by the AF predicted the extent to which WM-connected epileptic sources were influenced during delivery of direct stimulation therapy (Fig 1).

Conclusions: The preimplant AF-based model offers the potential to predict optimal implant sites for 2 4-contact depth leads interfacing with WM influencing up to 3 distant communicating epileptogenic sources. The AF takes into account patient-specific axon bundle orientation near the electrode contacts. A workflow incorporating such a model can potentially increase the number of patient candidates for RNS & DBS therapies.

Funding: Please list any funding that was received in support of this abstract.: Foglia Family Foundation, Mary Keane Fund.