Tesi etd-01202020-122307
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
CRACCHIOLO, MARINA
URN
etd-01202020-122307
Titolo
Neural decoding algorithms for bioelectronic medicine and neuroprosthetics
Settore scientifico disciplinare
Istituto di Biorobotica
Corso di studi
Istituto di Biorobotica - BIOROBOTICS
Commissione
relatore MICERA, SILVESTRO
Tutor Prof. MAZZONI, ALBERTO
Tutor Prof. MAZZONI, ALBERTO
Parole chiave
- neural decoding
- neuroprosthesis
- signal processing
Data inizio appello
08/05/2020;
Disponibilità
completa
Riassunto analitico
This thesis aims at advancing the field of neural recording in the peripheral nervous system for therapeutic purposes. The possibility to record and process neural signals is particularly appealing since it could allow to design more effective technological solutions to improve the quality of life of people with different types of disabilities.
In this thesis, we developed decoding frameworks to investigate complex physiological mechanisms and to restore lost physiological functions using neural pattern-based strategies.
The first two studies presented in this thesis fall within the area of bioelectronics medicine, an emerging field promising to treat chronic diseases by modulating autonomic nerves. Both these works helped to characterize the physiological system under investigation and addressed the lack of the knowledge necessary to optimize therapeutic strategies. The aim of the first work was to address metabolic disorders and in particular we focused on the activity of the carotid sinus nerve. Neural markers were extracted in healthy and diabetic rats to decode the metabolic status from the activity recorded in response to physiological stimuli. In the second study, electrical stimulation of the vagus nerve was used to investigate fiber types known to be involved in specific physiological mechanisms. Single fiber type activation was identified from the evoked compound neural activity, and a model for each fiber was built using stimulation parameters and physiological variables measured with non-invasive methods.
The second part of the thesis is dedicated to the development of a framework to decode motor intention from peripheral intraneural recordings in an upper limb amputee aimed to control a hand prosthesis. For the first time, TIMEs were used in the somatic nerve of a human subject for recording. We proposed a novel approach to process neurograms and reduce data dimension, leading to the decoding of 11 classes with high performance. Our strategy resulted to be stable and robust across days, envisioning a more natural and intuitive new generation of hand prostheses.
Across chapters, the level of complexity of both the setup and the decoding techniques increases. We went from preclinical and acute protocols in small animals (rats) with single-channel electrode, hook placed in the carotid and the cuff-like electrode in the vagus, to the implantation of four intraneural multichannel electrode (14x4 active sites) implanted chronically in the median and ulnar nerve of a human subject.
This thesis investigated the possibility to record peripheral neural activity for three applications in the autonomic and the somatic nervous systems. The results are very promising and pave the way for the development of complex and optimized system, working in closed-loop, managing both stimulation and recording.
In this thesis, we developed decoding frameworks to investigate complex physiological mechanisms and to restore lost physiological functions using neural pattern-based strategies.
The first two studies presented in this thesis fall within the area of bioelectronics medicine, an emerging field promising to treat chronic diseases by modulating autonomic nerves. Both these works helped to characterize the physiological system under investigation and addressed the lack of the knowledge necessary to optimize therapeutic strategies. The aim of the first work was to address metabolic disorders and in particular we focused on the activity of the carotid sinus nerve. Neural markers were extracted in healthy and diabetic rats to decode the metabolic status from the activity recorded in response to physiological stimuli. In the second study, electrical stimulation of the vagus nerve was used to investigate fiber types known to be involved in specific physiological mechanisms. Single fiber type activation was identified from the evoked compound neural activity, and a model for each fiber was built using stimulation parameters and physiological variables measured with non-invasive methods.
The second part of the thesis is dedicated to the development of a framework to decode motor intention from peripheral intraneural recordings in an upper limb amputee aimed to control a hand prosthesis. For the first time, TIMEs were used in the somatic nerve of a human subject for recording. We proposed a novel approach to process neurograms and reduce data dimension, leading to the decoding of 11 classes with high performance. Our strategy resulted to be stable and robust across days, envisioning a more natural and intuitive new generation of hand prostheses.
Across chapters, the level of complexity of both the setup and the decoding techniques increases. We went from preclinical and acute protocols in small animals (rats) with single-channel electrode, hook placed in the carotid and the cuff-like electrode in the vagus, to the implantation of four intraneural multichannel electrode (14x4 active sites) implanted chronically in the median and ulnar nerve of a human subject.
This thesis investigated the possibility to record peripheral neural activity for three applications in the autonomic and the somatic nervous systems. The results are very promising and pave the way for the development of complex and optimized system, working in closed-loop, managing both stimulation and recording.
File
| Nome file | Dimensione |
|---|---|
| Thesis_P...arina.pdf | 6.18 Mb |
Contatta l'autore |
|