For the past few years, interest in non-surgical and temporarily implanted brain stimulation methods as tools for clinical studies and research has grown substantially.
In electrocorticography (ECoG) stimulation, an array of electrodes is placed on the cortical surface during a surgical procedure, most typically as part of surgical planning for resection of epileptogenic tissue. These ECoG electrodes can be used to both measure intrinsic brain activity and to stimulate superficial cortical regions. Non-invasive electrical stimulation achieved through electrodes placed on the scalp, generally referred to as trancranial Current Stimulation (tCS), is frequently used to stimulate superficial brain areas. This thesis explores some questions related to modeling and optimization of ECoG and tCS stimulation. Both topics depend on computational modeling of the distribution of current in the head induced by the stimulation, typically carried out use the Finite Element Method (FEM). FEM models have been validated for tCS but not for ECoG stimulation.
In the first part of this project, as part of a larger collaboration, we analyze ECoG data recorded with arrays of electrodes during stimulation episodes and compare the results of our analysis to FEM simulations as an initial attempt to quantify accuracy of the FEM modeling.
In the second part of the thesis we focus on a recently published method that offers promise to allow deeper and more focused tCS than has been possible to date. Specifically, in a very recent study, Grossman and co-workers described a non-invasive method to stimulate deep brain regions that they called Temporal interference (TI) stimulation. The intuition behind this work is that, since neurons respond only to relatively low frequencies (typically below a few hundred Hertz), by applying multiple different oscillating currents at nearby frequencies above that range, the superposition of these currents can create a low frequency envelope electric field in deep areas of the brain with a large enough magnitude to modulate neuron firing in deep brain regions while not stimulating overlying regions. This work raises many questions regarding electrode configuration and current injection patterns that might best stimulate a specific desired region of interest.
In this thesis, we investigate the use of a multi- electrode FEM-based optimization approach to maximize the TI effect in a region of interest while limiting the TI effect in non-targeted regions. We study this problem applying FEM in a spherical simulation model with multiple target regions. We analyze the relevant objective function, which turns out to be non-convex but in a highly structured manner, and formulate and study via numerical simulations a constrained optimization problem based on that objective. Results indicate some ability to control delivery of TI stimulation but several aspects of this optimization are not yet well understood.
- Professor Dana Brooks (Advisor)
- Professor Stratis Ioannidis
- Dr. Sumientra Rampersad