Brain-spine interface for the volitional control of trans-spinal magnetic stimulation to rehabilitate movement in paralyzed patients

Abstract

Neurological pathologies damaging communication between central nervous system (CNS) and peripheral nervous system (PNS), such as stroke, can result in permanent motor impairment. Around 20% of stroke patients making a partial recovery remain with foot drop, which is expressed as weakness of ankle joint and abnormal muscle activations and movement coordination during locomotion. For these patients, the accident has an enormous impact in their life quality, social integration and economic situation. Currently, there is no effective treatment for this pathology and, given the rising worldwide prevalence of stroke during the upcoming years, there is a great need of improving existing rehabilitative interventions.

In this line, much research has focused on developing technology to improve rehabilitation and motor restoration of paralyzed patients. Advances in technology have considerably contributed to the capacity of acquiring, decoding and manipulating neural activity, and has been applied in clinical environments demonstrating its rehabilitative efficacy. Spinal cord stimulation (SCS) has emerged as a powerful tool for manipulation the spinal circuitry to generate walking-like patterns and to regain partial motor control of paralyzed limbs. Comparing to passive SCS, using brain activity to activate the stimulation is a more natural approach and allows volitional control of it. Brain-spine interfaces (BSI) have appeared as a technology to acquire, process and transform neural signals into commands to control SCS. BSIs enable associative connection between brain neural activity encoding motor intentions and activation of spinal pools and muscles, which may trigger Hebbian mechanisms that favor neuroplasticity boosting rehabilitative effects. To date, only implantable BSIs have been devised and tested in animals. Developing non-invasive BSIs would broaden the field of applicability of this rehabilitative technology on patients with motor disorders.

This thesis works on the development of a non-invasive BSI, based on continuous control of trans-spinal magnetic stimulation (ts-MS) using electroencephalographic activity (EEG), for motor rehabilitation of patients with lower limb paralysis. To this end, a set of 6 studies is presented, addressing the challenges and questions from three lines of work: (1) relevance of artifact removal methods, (2) neuromodulation by electromagnetic stimulation, and (3) conception of a brain-spine interface for motor rehabilitation.

Firstly, towards integrating neurostimulators in closed-loop systems, we characterize how artifacts of electromagnetic and neurophysiological origin interfere with electrophysiological recordings reflecting active participation of the patient and simulate their impact in closed-loop control. Contamination of electrophysiological recordings hampers the estimation of neural activity of interest and directly affects the performance of rehabilitative systems. Particularly for brain-controlled neural interfaces, we propose a median filter to minimize stimulation artifacts and statistical thresholding to eliminate neurophysiological artifacts. We demonstrate the need of adequate artifact removal methods that avoid bias in the quantification of task-related motor activity allowing contingent peripheral feedback.

Secondly, we investigate the modulatory effects of electromagnetic stimulators on the nervous system at different levels and discuss its implications in neurorehabilitation. Two studies are proposed: one evaluates the influence of neuromuscular electrical stimulation (NMES) on sensorimotor excitability, and one investigates the influence of ts-MS on cortico-spino-muscular excitability. We evidence that intensity and dose of NMES can produce immediate and cumulative effects on sensorimotor excitability measured with EEG. Our exploratory study shows that lumbar ts-MS strengthens the corticomuscular efficacy. Understanding how passive stimulation interacts with the nervous system could help us to improve interventions based on closed-loop stimulation.

The third line of work composing this thesis focuses on the development and validation of a BSI for lower-limb motor rehabilitation. Given the knowledge acquired from the other two lines of work, we aim at engineering the first non-invasive BSI that integrates ts-MS controlled by EEG activity. We prove the effectiveness and robustness of the system to work in real-time eliminating stimulation artifacts and engaging motor and sensory nervous system. We test and validate the BSI system in healthy participants and stroke patients and measure the neurophysiological, clinical and behavioral changes induced by an intervention based on our system. Our results show that the BSI increases the excitability of corticomuscular pathways, improves sensorimotor function, enhances balance and walking speed, and decreases spasticity.

In summary, this thesis proposes a novel non-invasive BSI that enables contingent connection between CNS and PNS, and evaluates its feasibility for rehabilitation of patients with motor paralysis. The here presented system provide relevant insights towards designing and developing of this innovative technology and may set the basis for future new investigations.