| Author: | Zhang, Jianing |
| Title: | Trans-spinal electrical stimulation-induced sensorimotor rehabilitation and EEG-based machine learning evaluation for upper extremity after stroke |
| Advisors: | Hu, Xiaoling (BME) Zhou, Yongjin (SO) |
| Degree: | Ph.D. |
| Year: | 2024 |
| Subject: | Cerebrovascular disease -- Patients -- Rehabilitation Arm -- Innervation Neural stimulation Electrotherapeutics Electroencephalography Machine learning Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Biomedical Engineering |
| Pages: | xxiv, 166 pages : color illustrations |
| Language: | English |
| Abstract: | Upper extremity (UE) sensorimotor impairments are a significant cause of post-stroke long-term disability. Sensory deficits can impact motor outcomes and hinder participation in daily activities. However, sensory impairments are often ignored in traditional practices because of the lack of reliable measures. Manual measurements depend on subjective experiences, which are hard to maintain consistently across a larger stroke population. To address this issue, electroencephalography (EEG) has been used to identify transient sensory neural responses and provide objective data for sensory impairments. However, its interpretation still relies heavily on human professionals, a process that can be both time-consuming and labor-intensive given the large amount of data generated. Machine-learning (ML) techniques, specifically support vector machine (SVM) models with kernel functions, can help reduce the burden of analyzing neuroimaging data. These models can automatically analyze massive amounts of data and make predictions. However, the automatic evaluation of EEG data in post-stroke sensory impairments using SVM techniques is yet to be fully investigated. In addition to the sensorimotor evaluation, conventional physical training is the usual therapy for motor recovery after stroke, along with sensorimotor evaluation. These therapies require intensive and repeated exercises to improve sensorimotor function. However, they do not produce significant long-term results. This may be due to the inadequate central nervous system (CNS) stimulation for neuronal changes. Noninvasive stimulation of the spinal cord, such as trans-spinal electrical stimulation (tsES), aims to increase the excitability of the spinal circuits and the responsiveness of the remaining neural pathways. Some studies have shown that spinal cord electrical stimulation can enhance upper limb motor control and decrease muscle spasticity after stroke. However, more research is needed to assess the immediate effect of tsES on the cortical signals that control the peripheral muscles during voluntary movements of the UE after stroke. Also, the rehabilitation effects of tsES on the interactions of cortical, spinal, and muscle activities after stroke are poorly understood. Therefore, the main objectives of this study were: (i) to establish an EEG-based SVM classification model to evaluate poststroke impairments in fine tactile sensation automatically; (ii) to evaluate the immediate effects of tsES on the cortical and muscular signals during voluntary UE contractions; (iii) to investigate the rehabilitation effects of tsES and voluntary physical training on the interactions of cortical, spinal, and muscular signals during upper limb movements in the long-term. The study was conducted as follows: The first section developed an ML model incorporating SVM to assess post-stroke impairments related to fine tactile sensation. The experiment involved stroke and unimpaired participants. Stimulations were administered using cotton, nylon, and wool fabrics, targeting different UE of stroke participants and the dominant UE of unimpaired participants. The average and maximal relative spectral power (RSP) values of the EEG signals were utilized as inputs to feed the SVM model. The model's generalization exhibited noteworthy accuracy variances when evaluating fabric stimulations within higher frequency bands, specifically the beta/gamma range. The EEG-based SVM-ML model aligned with the manual assessment of cortical responses to textile stimulations, indicating its potential for the automatic evaluation of fine tactile sensations following a stroke. The second section examined the immediate effects of tsES on the cortical and muscular signals during voluntary UE contractions in chronic stroke patients. Twelve patients performed wrist-hand motion tasks at submaximal levels with tsES applied to the cervical spinal cord. Data acquisition involved collecting both EEG and EMG data from the sensorimotor cortex and the distal and proximal muscles of the UE. The cortico-muscular coherence (CMCoh), laterality index (LI) of peak CMCoh, and EMG activation level parameters were compared between non-tsES and tsES conditions. The results showed that tsES significantly increased the CMCoh and LI in the agonist distal muscles, decreased the activation levels of EMG in the antagonist distal muscle and proximal UE muscles, and increased the LI of the proximal UE muscles. The third section investigated how cervical tsES training affects the patterns of cortico-muscular descending signals during voluntary movements in chronic stroke patients. Twenty patients were divided into tsES and control groups. They underwent twenty sessions of tsES with VPT or VPT alone. The evaluation outcomes, including clinical scores, CMCoh, LI, and EMG activation level, were measured before, after, and three months after the training. The tsES group showed significant differences in the outcomes across the sessions. The clinical scores, such as FMA and MAS, improved significantly. The laterality index of distal and proximal muscles increased significantly. The CMCoh and EMG activation levels of antagonist distal and proximal muscles decreased significantly. In conclusion, the EEG-based SVM-ML model exhibited outcomes that closely resembled the manual assessment of cortical responses to fabric stimulations; this could help to automate the measurement of fine tactile sensations in individuals who have experienced a stroke. In addition, the non-invasive cervical tsES combined with VPT in chronic stroke patients enhanced upper limb functional outcomes and reduced muscular spasticity. It also enhanced the responsiveness of residual descending pathways by increasing spinal cord excitability while reducing compensatory effects in proximal upper limb muscles. These findings suggested that tsES could be used as an adjunct to physical rehabilitation to facilitate long-term recovery of upper limb motor function in individuals with chronic stroke. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13096