Experimental evaluation of the impact of sEMG interfaces in enhancing embodiment of virtual myoelectric prostheses

Abstract

Introduction: Despite recent technological advances that have led to sophisticated bionic prostheses, attaining embodied solutions still remains a challenge. Recently, the investigation of prosthetic embodiment has become a topic of interest in the research community, which deals with enhancing the perception of artificial limbs as part of users' own body. Surface electromyography (sEMG) interfaces have emerged as a promising technology for enhancing upper-limb prosthetic control. However, little is known about the impact of these sEMG interfaces on users' experience regarding embodiment and their interaction with different functional levels.

Methods: To investigate this aspect, a comparison is conducted among sEMG configurations with different number of sensors (4 and 16 channels) and different time delay. We used a regression algorithm to simultaneously control hand closing/opening and forearm pronation/supination in an immersive virtual reality environment. The experimental evaluation includes 24 able-bodied subjects and one prosthesis user. We assess functionality with the Target Achievement Control test, and the sense of embodiment with a metric for the users perception of self-location, together with a standard survey.

Results: Among the four tested conditions, results proved a higher subjective embodiment when participants used sEMG interfaces employing an increased number of sensors. Regarding functionality, significant improvement over time is observed in the same conditions, independently of the time delay implemented.

Conclusions: Our work indicates that a sufficient number of sEMG sensors improves both, functional and subjective embodiment outcomes. This prompts discussion regarding the potential relationship between these two aspects present in bionic integration. Similar embodiment outcomes are observed in the prosthesis user, showing also differences due to the time delay, and demonstrating the influence of sEMG interfaces on the sense of agency.

Keywords: Embodiment; Surface EMG; Upper-limb prosthesis; Virtual reality.

Fig. 1

The figure shows the different sEMG configurations. The low density configuration uses channels number 1 and 5 of the first Myo armband and channels 3 and 7 from the second Myo armband. The high density condition uses all 8 channels from both Myo armbands. The alignment of the armbands on the arm can be seen on the right, for the example of a left handed person. All participants used their dominant hand during the experiments

Fig. 2

The figure shows a schematic of the experimental protocol. After giving their informed consent, the experiment starts with the sEMG data recording, and model training. The first part of the experiment focuses on the embodiment evaluation, conducting a baseline measurement, a familiarization phase in a gaming environment and a second embodiment measurment. This part of the experiment was repeated for the four conditions. After a short break, participants continue the training and functionalities assessment only with the synchronous conditions (half of them with condition C2 while the other half with condition C4)

Fig. 3

Experimental setup used in the study. The figure shows a participant wearing the HTC Vive Pro head mounted display (HMD), used to visualize the virtual environment. Two Myo armbands (up to 8 sEMG sensors each) and one HTC Vive tracker are placed on the user’s forearm

Fig. 4

The figure presents the gaming environment which was designed and used in the familiarization phase of this study. The environment consists of a shooting game, where the user can grasp one of the three water pistols placed on the table and shoot only water balloons with a matching color. In the picture, a participant is holding a blue water pistol and pointing it toward a falling water balloon to break it

Fig. 5

Schematic visualization of the proprioceptive drift (PPD) (a) and the virtual proprioception error (VPE) (b). The PPD is computed over the difference from the real hands position ($P_{\mathit{rh}}$) and where the participant thinks the real hand is ($P_{\mathit{prh}}$). The VPE computes the difference between the virtual hand ($P_{\mathit{vh}}$) and where the participant thinks the virtual hand is ($P_{\mathit{pvh}}$)

Fig. 6

The pictures show the five different virtual training environments designed and used in this study. a Opening a door by grasping and turning a doorhandle. b Holding a key to be inserted into a lock. c Puring balls from a cup into a bowl. d “Clean Sweep” task, known from Cybathlon 2020 [40]. e Box and blocks environment, where users can pick and place red blocks from one box to the other

Fig. 7

Virtual arm proprioception error (VPE). Red diamonds present outcomes from the prosthesis user, excluded from the statistical analysis and solely used for representing the target population. a Shows the ${\text{VPE}}_C$ for the baseline and the conditions without taking the baseline into account (Eq. 1). In b we visualize the VPE according to Eq. 2. Estimated means and standard errors from a post-hoc Tukey test are reported in a barplot format. The data was transformed via a quantile transformation for the statistical analysis, and its inverse transform was applied to the results presented

Fig. 8

Subjective embodiment results from the questionnaire. a Displays the distribution of the experimental data, separated into the three components of the survey: ownership, location and agency. Data is presented in violin plots with their respective mean. Red diamonds reports results from the prosthesis user, excluded from the statistical analysis and solely used for representing the target population. b displays the estimated means and standard errors from a post-hoc Tukey test, with asterisks that mark significance between components, together with the Cohens distance d. c reports the estimated means and standard errors from a post-hoc Tukey test, with asterisks that mark significance between tested conditions. The lower the score, the stronger the participants agreed with the items from Table 1, which indicates a stronger embodiment

Fig. 9

Functional results from the target achievement control (TAC) test. TAC metrics are reported in row-wise groups: Success rate in a–c, time to target in d–f, and accumulative time in target in g–i. For all metrics, column-wise groups present results for different factors and their interactions: panels (a, d, g) shows the results for Sensors: conditions (C2) vs (C4); panels (b, e, h) for Training; and panels (c,f,i) for Sensors x Training. All panels display the estimated means and standard errors from a post-hoc Tukey test, with asterisks that mark significance between conditions, together with the Cohens distance d. LD = low density; HD = high density; pre = pre-training; post = post-training

Fig. 10

Correlation between subjective embodiment and functionality metric. TAC Success Rate is selected as most representative functional metric for this study. Note that the TAC values correspond to the Pre training session, conducted immediately after participants provided their responses regarding the subjective embodiment following the Interaction Phase with the tested controller (see Fig. 1). Values proximal to the regression line indicate the slope of their relationship