Luigi Borda
AI-based optimization of somatosensory neuroprosthetic stimulation through learning control algorithms.
Rel. Danilo Demarchi. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2021
Abstract: |
Millions of people worldwide suffer from peripheral neuropathy, which damages the peripheral nervous system with a devastating impact on the quality of life. Peripheral neuropathy causes reduced peripheral sensitivity of the affected limbs and hence impacts the control of proprioceptive feedback during locomotion. Being such a widespread problem, in recent years researchers are looking for techniques to artificially restore sensory feedback to people with neuropathy. One of the most innovative solution consists in the use of TENS (Transcutaneous Electrical Nerve Stimulation) to stimulate non-invasively the peripheral nerves. Although encouraging results to restore sensory feedback have been obtained using this technique, the choice of stimulation parameters remains a very topical problem given the multidimensional space of possible parameters to be explored before choosing the optimal setting. The goal of my thesis is therefore to develop a closed-loop system based on AI-algorithm that allows an automatic and efficient identification of stimulation parameters for neuroprosthetic applications. To achieve this, I developed a VR-TENS platform in which the patient is immersed in a virtual environment (VR) and provided with nerve stimulation by means of TENS. The electrically evoked sensations are collected by the VR-TENS platform and an AI-algorithm based on reinforcement learning will identify the optimal stimulation parameters in order to elicit a comfortable, intuitive and functional artificial sensation. The first chapter introduces the problem of peripheral neuropathy, the state of the art of techniques to re-establish sensory feedback and presents a general overview of peripheral nerve stimulation. In this chapter, I also describe the state of the art of the relevant AI-algorithms allowing an automatic and efficient identification of stimulation parameters and I then introduce reinforcement learning. In the second chapter, I describe the overall VR-TENS system in details: the TENS paradigm for sensory feedback, the virtual reality environment and the AI-algorithm based on reinforcement learning. Then, I present how to integrate the different technologies to create the AI-based closed-loop platform for neuroprosthetic application. Finally, I report the results obtained by testing and validating the overall system on healthy subjects. The proposed system therefore represents a breakthrough in the challenging field of artificial sensory feedback restoration techniques, by providing aid to clinicians with an automatic and easy-to-use platform. |
---|---|
Relators: | Danilo Demarchi |
Academic year: | 2020/21 |
Publication type: | Electronic |
Number of Pages: | 148 |
Additional Information: | Tesi secretata. Full text non presente |
Subjects: | |
Corso di laurea: | Corso di laurea magistrale in Ingegneria Biomedica |
Classe di laurea: | New organization > Master science > LM-21 - BIOMEDICAL ENGINEERING |
Ente in cotutela: | ETH- Neuroengineering lab (SVIZZERA) |
Aziende collaboratrici: | ETH Zurich |
URI: | http://webthesis.biblio.polito.it/id/eprint/17534 |
Modify record (reserved for operators) |