Date on Master's Thesis/Doctoral Dissertation


Document Type

Doctoral Dissertation

Degree Name

Ph. D.


Electrical and Computer Engineering

Degree Program

Electrical Engineering, PhD

Committee Chair

Naber, John

Committee Co-Chair (if applicable)

Keynton, Robert

Committee Member

Keynton, Robert

Committee Member

Walsh, Kevin

Committee Member

Zurada, Jacek

Committee Member

Roussel, Thomas

Author's Keywords

neuron; simulation; genetic algorithm; spinal cord stimulation; pulse position optimization; electric field modeling


The use of neural prostheses to improve health of paraplegics has been a prime interest of neuroscientists over the last few decades. Scientists have performed experiments with spinal cord stimulation (SCS) to enable voluntary motor function of paralyzed patients. However, the experimentation on the human spinal cord is not a trivial task. Therefore, modeling and simulation techniques play a significant role in understanding the underlying concepts and mechanics of the spinal cord stimulation. In this work, simulation and modeling techniques related to spinal cord stimulation were investigated. The initial work was intended to visualize the electric field distribution patterns in the spinal cord. A system consisting a set of stimulating electrodes with a model of the spinal cord was built and simulations were performed by applying different stimuli to the electrodes. Depending on the complexity of the model, the simulation times can be varying. However, the system demonstrated the ability to visualize the field distributions inside the spinal cord for static and transient stimuli. This data can be used to aid experimental studies and to better understand the results of the spinal cord stimulation by mapping known experimental results to the simulation results. The stimulation of biological tissue raises safety concerns. Therefore, a methodology utilizes in electrical stimulation of biological systems known as charge balancing was studied. The importance of charge balancing and the effects of using charge balancing was studied with simulation studies. The simulation results showed extensive charge accumulation in the biological tissue, around the stimulation sites, when charge balancing was not utilized. Hence, the studies showed the importance of using charge balancing and the importance of avoiding pulse collisions. Stimulation sequences that utilizes different stimuli with different frequencies results in a phenomenon known as pulse collision. Pulse collisions cause complications in charge balancing and should be avoided or reduced in stimulation procedures. An algorithm and a software based on the algorithm was developed to reduce pulse collisions by optimizing pulse positioning. The tool showed significant reduction in pulse collisions for simulation sequences that were used in experimental settings. The current version of the software is capable of optimizing up to five different pulses. Coupling the electric fields and action potentials generated in neurons due to the electric fields can lead to significant discoveries on the mechanics of the SCS. Therefore, the investigation of bi-domain models has valuable implications. Bi-domain models were built by combining different neural models with electric field results. The simulations proved the possibility of visualizing the action potentials in neurons, when the spinal cord was stimulated using electrical pulses. Having the optimized neural models that simulate the behavior of the motor neurons can aid the experimental studies by easily identifying the proper stimulation sequences to activate motor neurons. The final section of this work was focused on using machine learning in neuroscience for optimization purposes. Well known Hodgkin-Huxley (HH) model and a recently published Izhikevich mathematical model were used for this task. The HH model is a widely used, biophysically meaningful model that can simulate the action potentials in the nerve axons. It is mainly used to simulate the type 2 behavior of the axon firing. However, other types of neuron spiking and bursting have been observed in the literature. This work demonstrates the optimization of the HH model parameters to simulate neuron bursting behavior. The results of the study demonstrate that it is possible to extend the HH model beyond its intended type 2 behavior and can be modified to simulate more complex neuron firing patterns including neuron bursting. The optimized HH model was able to generate bursting patterns corresponding to the two target bursting patterns used in this study with error values