The effect of HD sEMG-based Augmented Reality visual Feedback on Motor Performance during a Submaximal Isometric Knee Extension Endurance Task.

Biofeedback technique consists in the real-time provision of physiology-related information acquired from an individual, as an extension of his intrinsic sensorial feedback. The use of EMG-based biofeedback in sport and rehabilitation contexts has been widely investigated at research level and nowadays it is considered having a great potential in influencing voluntary muscle activation. Consequently, it may be used as a promising treatment for a broad range of musculoskeletal conditions. Traditionally, biofeedback has been delivered in real-time through simple visual indicators, acoustic, or vibrotactile stimuli. The recent development of Augmented Reality (AR) technologies constituted the technological premise allowing the use of this technique in the sEMG biofeedback field. This master thesis project aims at investigating the impact of using an AR-based visual biofeedback driven by High Density surface Electromyographic (HD-sEMG) signals, on motor performance during an isometric knee extension task. The HD-sEMG/AR systems have been developed by LISiN, Italy. The study was performed in strict collaboration with the School of Sport, Exercise, and Rehabilitation Sciences at the University of Birmingham. 23 healthy individuals and 9 diagnosed with Patellofemoral Pain Syndrome were recruited. Both groups underwent 2 randomized measurement sessions: one performing an isometric knee extension while using a real-time AR visual feedback, and one without. During the fatiguing contraction with the AR, the participant was required to modify his muscles activity, alternating the activation zones between Vastus Medialis and Lateralis. HD-sEMG signals were collected by means of two 8x4 electrodes grids placed on the target muscles. The real-time RMS maps were computed, codified in a color scale, and provided both to the subject wearing smartglasses, and to the operator, monitoring the task correct execution, through a tablet. An additional visual and auditory feedback driven by a dynamometric signal was showed on a screen to maintain a 30% MVC constant torque during the tasks. Data analysis focused on computing the endurance time for each participant’s session using an automatic segmentation technique. The statistical analysis revealed significant differences between groups in terms of longer endurance time for the asymptomatic group and, inside this group, an increase in the endurance time in the session with AR biofeedback compared to the one without. A standard muscle fatigue analysis was conducted to determine if this physiological process may have been influenced by the use of AR. However, no statistical differences were highlighted in the comparison between the RMS and MNF normalized slopes during sessions with and without AR in both groups. Spatio-temporal distribution of myoelectric activity was studied computing the shift of the EMG signals’ amplitude centroid over time. Asymptomatic group displayed a larger shift in both vasti along the medio-lateral axis during session with AR, while symptomatic group exhibited the opposite trend. In summary, the results don’t show significant differences in muscle activation between the two sessions. However, an increase in endurance time is observed with AR. This leads to the hypothesis of a potential difference in muscle activation that has not been evident from the conducted analyses. Further investigations and technological developments are imperative to comprehensively assess its potential in enhancing motor task precision and performance.