Please use this identifier to cite or link to this item: https://hdl.handle.net/11000/35609

Interfaz cerebro-máquina multiparadigma para el control de la marcha con un exoesqueleto de miembro inferior


Thumbnail

View/Open:
 Tesis_Doctoral (SF) Laura Ferrero Montes.pdf

48,19 MB
Adobe PDF
Share:
Title:
Interfaz cerebro-máquina multiparadigma para el control de la marcha con un exoesqueleto de miembro inferior
Authors:
Ferrero Montes, Laura
Tutor:
Azorín Poveda, José María
Ortíz García, Mario
Editor:
Universidad Miguel Hérnández de Elche
Department:
Departamentos de la UMH::Ingeniería de Sistemas y Automática
Issue Date:
2024
URI:
https://hdl.handle.net/11000/35609
Abstract:
La lesión de la médula espinal (LME) ocurre como resultado de un daño en la médula que interrumpe las vías sensoriales y motoras, lo que puede llevar a una alteración temporal o permanente de la función motora y/o sensitiva. Cuando la lesión afecta a los miembros inferiores, repercute en la marcha,...  Ver más
Spinal cord injury (SCI) occurs when damage to the spinal cord disrupts sensory and motor pathways, resulting in temporary or permanent impairment of motor and/or sensory function. When the injury affects the lower limbs, it affects gait, leading to potential stigmatization and increased vulnerability to further injuries. Moreover, the loss or deterioration of the ability to walk independently impacts autonomy and the sense of freedom, ultimately diminishing the overall quality of life. The majority of SCI injuries are incomplete, meaning that there may be preserved spinal tissue that can be rehabilitated and repaired to restore motor function. Conventional therapies typically involve repetitive specific activities performed with the assistance of physiotherapists. Evidence supports the notion that the quantity and specificity of practice play crucial roles in promoting recovery and the development of motor skills. Recent advancements in robotic devices have facilitated their integration into treatments as an alternative to conventional therapies, requiring less effort from physiotherapists and providing extended training sessions with more reproducible patterns of movement. Various methods exist for controlling robotic devices, including mobile applications, joysticks, and command consoles. However, brain-machine interfaces (BMIs) present an alternative solution by enabling direct control through the mind. This technology holds promise for the development of innovative rehabilitation therapies, as users can mentally simulate specific movements, referred to as motor imagery (MI), to generate commands that translate into the movement of a robotic exoskeleton. From a rehabilitation perspective, MI promotes motor learning and, consequently, enhances the recovery process. This doctoral thesis primarily focuses on the design and evaluation of a BMI system aimed at controlling the initiation and cessation of lower limb exoskeleton gait based on users’ brain activity, specifically by detecting gait-related MI. The primary objective of this system is to provide assistance to individuals with reduced mobility and promote their rehabilitation progress. Within this context, three main lines of research have been pursued: the control of a lower limb robotic exoskeleton, the study of subjects’ brain patterns, and training within a virtual reality (VR) environment. The first line of research centers on the control of a lower limb robotic exoskeleton. The BMI system monitors users’ brain patterns in real-time, making decisions at regular 0.5- second intervals. Two distinct control approaches were tested. The first approach combined MI and attention, requiring users to imagine the act of walking to activate the exoskeleton and relax their mind to halt the movement, but only when they were highly focused on the task. This approach ensured safer control by monitoring users’ attention levels during testing. The second approach relied solely on MI without incorporating additional attention-related factors. The use of attention levels was disregarded due to the excessive constraints imposed by the BMI system, significantly impacting performance and causing frustration among the subjects. In addition to the interface design, training and evaluation protocols for the system were developed. Two alternatives were explored in this regard. The initial approach involved users practicing MI tasks while walking with the assistance of the exoskeleton and practicing relaxation while remaining stationary. Afterwards, for evaluation, a model trained with these two tasks was used to control the system. However, it was observed that many of the differences detected were a result of artifacts generated by movement, which masked the true brain signals. Consequently, a second approach was adopted, where users performed mental tasks under static conditions for half of the trials and, for the remaining half, performed the same tasks while the exoskeleton facilitated movement. Subsequently, during the system evaluation, two models were utilized: one trained with static trials and another trained with trials during movement, based on the kinematic state of the exoskeleton. This evaluation was conducted comprehensively, encompassing both healthy individuals and patients with spinal cord injuries. In the second line of research, patterns of brain activity generated in the electroencephalographic signals during Mi were studied in comparison to patterns generated during relaxation. Differences in these patterns among participants and across experimental sessions were also analyzed. As a result, significant differences in brain activity were observed both among participants and across sessions, suggesting the presence of learning. The third line of research incorporated training with VR to reduce BMI calibration time. Participants were immersed in a virtual environment where their imagined walking actions corresponded to the progression of an avatar. Training in this environment preceded assuming control of the exoskeleton via the BMI. Performance improvements and brain pattern analyses were conducted for a group that received VR training, comparing it to a control that received more extensive BMI training without VR. The results indicated superior performance by the VR group compared to the control group. In summary, this doctoral thesis has focused on the development and evaluation of a BMI for controlling a lower limb exoskeleton, emphasizing the importance of using MI as a fundamental control mechanism. Throughout this research, effective control strategies have been identified, and significant brain differences have been observed when comparing brain activity during MI with states of relaxation. Furthermore, it has been demonstrated that integrating VR into BMI training leads to improvements in user performance. These findings support the relevance and applicability of BMIs in the field of rehabilitation for individuals with reduced mobility, opening new perspectives to enhance their quality of life.
Keywords/Subjects:
Rehabilitación médica
Neurociencias
Tratamiento de señales
Knowledge area:
CDU: Ciencias aplicadas: Ingeniería. Tecnología
Type of document:
info:eu-repo/semantics/doctoralThesis
Access rights:
info:eu-repo/semantics/openAccess
Appears in Collections:
Tesis doctorales - Ciencias e Ingenierías



Creative Commons ???jsp.display-item.text9???