Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Closed-loop optimization of vagus nerve stimulation for functionally selective neuromodulation
Development of a closed-loop machine learning algorithm to extend the therapeutical uses of vagus nerve stimulation by improving the selectivity of electrical nerve stimulation, trained on a computational model and validated against experimental data.
Electrical stimulation of the peripheral nervous system has been shown to have several therapeutical applications. Some of those have been approved for clinical use, such as vagus nerve stimulation in the treatment of epilepsy and depression and sacral nerve stimulation for bladder dysfunctions. The potentiality of applying neurostimulation has been demonstrated for the treatment of many more diseases, but one of the main obstacles for the growth of this technology is the difficulty of obtaining the desired outcome with minimal collateral effects. A clear example where this has been encountered is in previous attempts of restoring sensory feedback in users of limb prostheses: eliciting the desired sensation, mediated by a specific subgroup of fibers, is seldom possible and even when achieved rarely stable in time. The parameters space for electrode design and stimulation policy is very high-dimensional, and the optimal parameters to obtain the desired outcome depend on the surgical placement of the electrodes, on inter-individual differences, and are not stable in time due for example to shifts of the electrode, foreign body reaction, and regrowth of fibers. A closed-loop optimization algorithm is required to adapt the stimulation to the functional outcome of a system which is dependent on unmeasurable factors and evolving in time. Computational models simulating the effect of electrical stimulation of the single nerve fibers have been employed extensively to aid in the test of electrode and stimulation policy design and choice. They represent an excellent platform to experiment new stimulation strategies for selective activation of the fibers of interest, and as a test bench for the development of an algorithm for closed-loop optimization of the stimulation parameters.
Electrical stimulation of the peripheral nervous system has been shown to have several therapeutical applications. Some of those have been approved for clinical use, such as vagus nerve stimulation in the treatment of epilepsy and depression and sacral nerve stimulation for bladder dysfunctions. The potentiality of applying neurostimulation has been demonstrated for the treatment of many more diseases, but one of the main obstacles for the growth of this technology is the difficulty of obtaining the desired outcome with minimal collateral effects. A clear example where this has been encountered is in previous attempts of restoring sensory feedback in users of limb prostheses: eliciting the desired sensation, mediated by a specific subgroup of fibers, is seldom possible and even when achieved rarely stable in time. The parameters space for electrode design and stimulation policy is very high-dimensional, and the optimal parameters to obtain the desired outcome depend on the surgical placement of the electrodes, on inter-individual differences, and are not stable in time due for example to shifts of the electrode, foreign body reaction, and regrowth of fibers. A closed-loop optimization algorithm is required to adapt the stimulation to the functional outcome of a system which is dependent on unmeasurable factors and evolving in time. Computational models simulating the effect of electrical stimulation of the single nerve fibers have been employed extensively to aid in the test of electrode and stimulation policy design and choice. They represent an excellent platform to experiment new stimulation strategies for selective activation of the fibers of interest, and as a test bench for the development of an algorithm for closed-loop optimization of the stimulation parameters.
The main goal of this project is the development of a closed-loop optimization algorithm to improve the therapeutical outcomes of vagus nerve stimulation, leveraging unconventional stimulation techniques. The system will be developed and trained on an existing computational framework and will produce recommendations of stimulation policies to obtain diverse therapeutical outcomes, which the candidate will validate on animal experimental results obtained by a partner institution.
**Recommendable skills:** MATLAB; COMSOL; Python; Machine Learning.
**Time effort:** Full time Master's thesis (typically 6 months).
The main goal of this project is the development of a closed-loop optimization algorithm to improve the therapeutical outcomes of vagus nerve stimulation, leveraging unconventional stimulation techniques. The system will be developed and trained on an existing computational framework and will produce recommendations of stimulation policies to obtain diverse therapeutical outcomes, which the candidate will validate on animal experimental results obtained by a partner institution.
**Time effort:** Full time Master's thesis (typically 6 months).
Federico Ciotti, PhD student, Neuroengineering Lab, ETH Zürich, Switzerland; federico.ciotti@hest.ethz.ch
Nataljia Katic, PhD student, Neuroengineering Lab & Institute Mihajlo Pupin, Email: natalija.katic@pupin.rs
Dr. Stanisa Raspopovic, Assistant Professor, Head of the Neuroengineering Lab, ETH Zürich, Switzerland; stanisa.raspopovic@hest.ethz.ch
Federico Ciotti, PhD student, Neuroengineering Lab, ETH Zürich, Switzerland; federico.ciotti@hest.ethz.ch Nataljia Katic, PhD student, Neuroengineering Lab & Institute Mihajlo Pupin, Email: natalija.katic@pupin.rs Dr. Stanisa Raspopovic, Assistant Professor, Head of the Neuroengineering Lab, ETH Zürich, Switzerland; stanisa.raspopovic@hest.ethz.ch