|Investigation of exo-neuro-musculo-skeleton with neural-network-based evaluation for ankle-foot rehabilitation after stroke
|Hu, Xiaoling (BME)
|Cerebrovascular disease -- Patients -- Rehabilitation
Hong Kong Polytechnic University -- Dissertations
|Department of Biomedical Engineering
|xxi, 152 pages : color illustrations
|Stroke is one of the primary causes of adult hemiplegia globally. Conventional motor recovery of the hemiparetic limb necessitates repeated and intensive training for stroke survivors. However, the current rehabilitation service for motor restoration after discharge from the hospital is insufficient, particularly for the ambulation ability. Although 60%-80% of stroke survivors can walk independently, most of them exhibit long-term gait disturbances, including high gait asymmetry, lower walking speed, inability to walk far, and being more likely to fall, which affect their mobility and integration into the community. Thus, more effective, and readily accessible rehabilitation services or methods are required to enhance the ambulation ability of chronic stroke survivors to improve their life quality. On the other hand, the evaluation of the training effects during neurorehabilitation is also a crucial issue, which is commonly conducted by a blinded assessor (e.g., professional physiotherapists). Clinical assessment is hard to obtain owing to the shorthanded situation in the current healthcare system, e.g., professional therapists. The surface electromyography (sEMG) signals driven quantitative and objective evaluations have been used to track the training effects, e.g., the co-contraction index (CI) of muscle pairs and activation level of individual muscle. However, these quantitative metrics are not available online and cannot be robustly correlated to clinical scores. The objectives of this study were: (1) development of a data-driven model involving sEMG for facilitating an objective and automated metric of training effects for poststroke rehabilitation assisted by robots, (2) development of an exo-neuro-musculo-skeleton ankle-foot system with balance sensing feedback (ENMS-BF) for motor recovery of the paralyzed lower extremity after stroke, and (3) investigation of the assistive capability and rehabilitation effects of the proposed ENMS-BF on chronic stroke survivors, with both face-to-face individual training and remote self-help paired training. This study was implemented in three sections as follows:
In the first part, we constructed a backpropagation neural network (BPNN) model with the sEMG signals as the driven data, which matched the mapping relationship between the sEMG characteristics and commonly utilized clinical scales, i.e., the Modified Ashworth Scale (MAS) and the Fugl-Meyer Assessment (FMA). Twenty-nine individuals with chronic stroke completed a robot-assisted upper limb rehabilitation program, with the sEMG signals collected before and after the 20-session intervention. There were significant correlations (P<0.001) between the manually assessed and mapped FMA and MAS scores, within the labelled data captured before and after the intervention. The results showed that the proposed sEMG-driven model based on BPNN enables the automated tracking of motor recovery for chronic stroke survivors and demonstrated the potential to be applied in automated assessment post-stroke.
In the second section, we developed a novel ENMS-BF driven by plantar pressures to assist gait training by dynamic correction of foot drop and foot inversion. The ENMS-BF can be worn unilaterally onto the paretic lower limb with a weight of 0.47 kg. It consists of a soft-and-rigid musculoskeletal combination, i.e., musculoskeleton, two-channel neuromuscular electrical stimulation (NMES), and a tactile vibrator. The properties of pressure-to-torque transmission of the musculoskeleton were measured quantitatively. The results showed that the ENMS-BF could effectively correct foot drop and foot inversion in the hemiparetic gait pattern.
In the third section, the feasibility and rehabilitative effects of the ENMS-BF-assisted gait training after stroke were evaluated. Twelve stroke survivors participated in the individual gait training with close supervision. Then, another 12 individuals with chronic stroke were recruited in self-help paired training based on a cyber physical social system (CPSS) for remote social links. The results indicated that the ENMS-BF assisted gait training was feasible and effective in improving the motor function, gait pattern, and plantar pressure of the paralyzed lower limb in both groups. The developed ENMS-BF combining with CPSS could effectively facilitate self-help gait training with remote management and peer support.
In conclusion, the developed sEMG-driven model based on BPNN could facilitate the automated assessment of motor function recovery post-stroke. The developed ENMS-BF could assist in ankle dorsiflexion and self-correction of foot inversion during gait training. The ENMS-BF-assisted individual gait training was effective for improvements of lower limb motor function, gait pattern, and plantar balance in the paralyzed limb post-stroke. Based on the CPSS, the ENMS-BF-assisted paired training could support and facilitate self-help rehabilitation with professional management and social links with peers remotely.
|All rights reserved
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: