Abstract
The objectives of this study were to test the feasibility of the developed waterproof wearable device with a Surface Electromyography (sEMG) sensor and Inertial Measurement Unit (IMU) sensor by (1) comparing the onset duration of sEMG recordings from maximal voluntary contractions (MVC), (2) comparing the acceleration of arm movement from IMU, and (3) observing the reproducibility of onset duration and acceleration from the developed device for bicep brachii (BB) muscle between on dry-land, and in aquatic environments. Five healthy males participated in two experimental protocols with the activity of BB muscle of the left and right arms. Using the sEMG of BB muscle, the intra-class correlation coefficient (ICC) and typical error (CV%) were calculated to determine the reproducibility and precision of onset duration and acceleration, respectively. In case of onset duration, no significant differences were observed between land and aquatic condition (p = 0.9–0.98), and high reliability (ICC = 0.93–0.98) and precision (CV% = 2.7–6.4%) were observed. In addition, acceleration data shows no significant differences between land and aquatic condition (p = 0.89–0.93), and high reliability (ICC = 0.9–0.97) and precision (CV% = 7.9–9.2%). These comparable sEMG and acceleration values in both dry-land and aquatic environment supports the suitability of the proposed wearable device for musculoskeletal monitoring during aquatic therapy and rehabilitation as the integrity of the sEMG and acceleration recordings maintained during aquatic activities.
Clinical Relevance
This study and relevant experiment demonstrate the feasibility of the developed wearable device to support clinicians and therapists for musculoskeletal monitoring during aquatic therapy and rehabilitation.
I. INTRODUCTION
The benefits of hydrostatic forces (buoyancy) and drag resistances particular to the aquatic medium have led therapists, researchers, and trainers to suggest aquatic rehabilitation [1]-[3]. The substantial variation in these physical characteristics from land-based activities has an impact on the human body’s physiology and biomechanics [1]. Additionally, aquatic rehabilitation is safe from the potential physical risks due to gravity and may lessen the likelihood of falling [3]. Aquatic therapy is a promising and widely used form of rehabilitation for those who are obese or have balance or neurological disorders because it provides a special environment that may help with motor recovery [1]. The main advantages of aquatic therapy are that it provides improved balance, mobility, and increased functional independence recovery from stroke [3]. Other benefits of aquatic therapy have been reported to include acceleration of the recovery from injury thereby reducing overall recovery time.
To take these advantages of aquatic therapy and to verify its effectiveness, previous research has investigated sEMG recordings of different muscles when doing various exercises in an aquatic environment [4]. To observe a difference in the mean %MVC of muscles between on-land and aquatic environments in different phases of exercise, sEMG signals were acquired from lower limb muscles during squatting in both of the environments [5], [6]. Moreover, to analyze the neuromuscular responses during the performance of a sit-to-stand (STS) task in aquatic and on-land, the sEMG system was put in a waterproof cover and hydro-gel (Ag/AgCl) electrodes were used in [7]. However, it is mentioned by [4], [8], [9] that Ag/AgCl electrodes needed to be waterproofed to use in aquatic environments otherwise the collected sEMG signals integrity get compromised. Although the study in [10] was conducted using commercially available wireless and waterproof EMG system, it didn’t demonstrate any comparative analysis between on-land and aquatic environments to test the signal integrity as the experiments are not based on any identical condition during exercises in two different environments. It is noticeable that most EMG sensors utilized in the most previous relevant research were not waterproofed [5], [7], [8], [11], so there was needed a waterproofing process separately such as covering with waterproofing tapes. However, it may affect the signal quality directly and lead to discomfort for users.
In this study, we used carbon black/polydimethylsiloxane (CB/PDMS) electrodes [8], [12], [13] for stable and robust sEMG data collection aquatic and developed waterproof and wireless wearable musculoskeletal monitoring device to represent the quantitative measurements of sEMG and IMU data during aquatic therapy and rehabilitation and to offer better wearability than existing aquatic data collection settings. We investigated the feasibility and reproducibility of the developed device with a comparison of the sEMG and IMU sensor recordings from BB muscles derived from movements in on-land and aquatic environments.
II. METHODS
A. Participants
5 healthy male subjects [mean ± SD]: age, 26.6 ± 2.97 yrs; height, 172.8 ± 9.0 cm; body mass, 69.9±7.2 kg agreed to participate. All were informed about the procedures and potential risks and gave their written informed consent to participate in the study. Institutional Review Board (IRB) approval was obtained from the University of Massachusetts Amherst IRB (#22010038) to carry out the experiments.
B. Device and electrodes
The device consists of three subunits operating in parallel— the IMU, sEMG sensor unit, and flash memory unit. The CB/PDMS electrodes, presented by [8], were used for collecting sEMG signals in aquatic environments. The device is waterproof and powered up by a 135mAh 3.7V Li-Polymer battery which is rechargeable wirelessly. The accelerometer data were stored in flash memory from the IMU sensor to observe the balance between the left and right sides of the body. Simultaneously, the sEMG data were stored in flash memory during the following activities. The collected data were used for further signal processing and statistical analysis.
C. Experimental procedure
The device was fastened with elastic belt worn around the subject’s bicep and the sensor node was positioned on the BB of the subject. There are two different experiment protocols, and each experiment protocol was conducted both on-land and in aquatic conditions in a same manner. In protocol 1, at the beginning of the experiment, subjects were asked to stand straight by keeping the elbow in full extension position and keeping the arms parallel with the body. Then they were asked to lift the forearm (flexion) with the palm in the upward direction till the full flexion and slowly lowered the forearm (extension) back to the straight position and parallel with the body. On the other hand, in protocol 2, subjects were asked to keep both arms open and parallel with the shoulder and close the forearm with the palm in the forward direction till the full flexion, and return the arm back to the original position and parallel with the shoulder till full extension. Each activity had 3, 6, and 9 repetitions (reps) in 5 sets, respectively. For each experiment, we began the recording with 5 minutes of data while the muscle was in relaxation, then the subject transitioned to a contraction stage, which consisted of isometric contraction for 30 s followed by relaxation for 5 mins. We followed the same time sequence for sEMG signal recording on both protocols. Prior to each test, subjects were asked to practice the requested activities until they were getting familiar with them.
D. Statistical Analysis
Muscle activation was measured with 1 kHz sampling rate during the experiment both on-land and in aquatic environment. All statistical calculations were computed using Matlab 2022a. Descriptive data were calculated for demographic data (age, height, and weight). The data were presented in the format of mean±standard deviation. Based on the feature sets of onset time and acceleration, the mean of sEMG activities of BB muscles were analyzed to compare on-land and aquatic environments, with the level of significance at p ≤ 0.05.
III. RESULTS
The graphical representation of the raw EMG signals in In Fig. 3, from BB muscles during an MVC in each activity for a single participant demonstrates that the developed device in this study correctly preserved the integrity of the sEMG recordings under all circumstances. Mean, standard deviation along with descriptive statistics, ICC, and CV% scores of onset duration of BB muscles of the participants at different protocols at different environments are summarized in Table I. Also, the acceleration of the arm movement during obtaining the sEMG for onset detection is presented in Table II. No significant differences in the onset duration of the two sides of the body were observed as the exercises were performed under no load condition. ICC and CV% ranged between r = 0.93–0.98 and 2.7–6.4%, respectively, between environments for the two protocols.
Fig. 3.
Signals collected during (a) 3 reps set, (b) 6 reps set, and (c) 9 reps set of the activity in on land and aquatic environment of protocol 1.
TABLE I.
Onset Duration Statistical Analysis
| Left Hand | Right Hand | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Set | Land | Aquatic | F a | p b | ICC c | CV %d | Land | Aquatic | F | p | ICC | CV % | |
| Protocol 1 | Rep 3 | 3.8 ± 0.45 | 3.6 ± 0.5 | 0.81 | 0.93 | 0.97 | 4.7 | 3.8 ± 0.45 | 3.6 ± 0.5 | 0.81 | 0.93 | 0.96 | 4.7 |
| Rep 6 | 3.2 ± 0.3 | 3.1 ± 0.34 | 0.78 | 0.94 | 0.97 | 2.8 | 3.2 ± 0.3 | 3.1 ± 0.36 | 0.78 | 0.94 | 0.97 | 2.8 | |
| Rep 9 | 3 ± 0.26 | 2.7 ± 0.3 | 0.75 | 0.9 | 0.93 | 6.4 | 3 ± 0.26 | 2.8 ± 0.3 | 0.75 | 0.9 | 0.93 | 5.9 | |
| Protocol 2 | Rep 3 | 3.8 ± 0.72 | 3.5 ± 0.8 | 0.81 | 0.93 | 0.97 | 5.8 | 3.8 ± 0.72 | 3.5 ± 0.8 | 0.81 | 0.93 | 0.97 | 5.8 |
| Rep 6 | 3.2 ± 0.48 | 3.1 ± 0.52 | 0.85 | 0.92 | 0.98 | 2.7 | 3.2 ± 0.47 | 3.1 ± 0.53 | 0.79 | 0.93 | 0.98 | 2.8 | |
| Rep 9 | 3 ± 0.34 | 2.8 ± 0.4 | 0.72 | 0.9 | 0.94 | 5.7 | 3 ± 0.34 | 2.8 ± 0.4 | 0.72 | 0.9 | 0.94 | 5.7 | |
The ratio of the variance of two sides of data. F is closer to 1 means it is not rejecting the null hypothesis at the 5% significance level.
The calculated significance for two data sets (on land and aquatic).
Measurement of sEMG reproducibility, which means how much it is following the trend of on land in aquatic.
Calculated by the ratio of standard deviation and mean to measure the precision.
TABLE II.
Acceleration of Arms from IMU Sensor Statistical Analysis
| Left Hand | Right Hand | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Set (Rep) |
Stage | Land (ms−2) |
Aquatic (ms−2) |
F | p | ICC | CV% | Land (ms−2) |
Aquatic (ms−2) |
F | p | ICC | CV % | |
| Protocol 1 | 3 | Flex | 1.96 ± 0.21 | 1.78 ± 0.22 | 0.91 | 0.89 | 0.97 | 8.9 | 1.96 ± 0.21 | 1.78 ± 0.22 | 0.91 | 0.89 | 0.97 | 8.9 |
| Ext | 1.97 ± 0.22 | 1.78 ± 0.22 | 1 | 0.9 | 0.93 | 8.6 | 1.97 ± 0.21 | 1.78 ± 0.22 | 0.91 | 0.9 | 0.93 | 8.6 | ||
| 6 | Flex | 2.01 ± 0.24 | 1.86 ± 0.24 | 1 | 0.91 | 0.96 | 8.3 | 2.01 ± 0.23 | 1.86 ± 0.24 | 0.92 | 0.91 | 0.96 | 8.3 | |
| Ext | 2.03 ± 0.23 | 1.87 ± 0.23 | 1 | 0.92 | 0.94 | 8.6 | 2.03 ± 0.23 | 1.87 ± 0.23 | 1 | 0.92 | 0.94 | 8.6 | ||
| 9 | Flex | 2.1± 0.25 | 1.95 ± 0.25 | 1 | 0.93 | 0.9 | 7.9 | 2.1 ± 0.24 | 1.9 ± 0.25 | 0.92 | 0.93 | 0.9 | 7.9 | |
| Ext | 2.12 ± 0.24 | 1.96 ± 0.25 | 0.92 | 0.91 | 0.91 | 8 | 2.12 ± 0.24 | 1.96 ± 0.24 | 1 | 0.91 | 0.91 | 8 | ||
| Protocol 2 | 3 | Flex | 1.92± 0.22 | 1.73 ± 0.23 | 0.91 | 0.89 | 0.97 | 9.1 | 1.92 ± 0.23 | 1.73 ± 0.24 | 0.92 | 0.89 | 0.97 | 9.1 |
| Ext | 1.92± 0.21 | 1.73 ± 0.22 | 0.91 | 0.92 | 0.93 | 8.5 | 1.92 ± 0.22 | 1.73 ± 0.23 | 0.91 | 0.92 | 0.93 | 8.5 | ||
| 6 | Flex | 1.98± 0.24 | 1.82 ± 0.25 | 0.92 | 0.91 | 0.96 | 8.1 | 1.98 ± 0.25 | 1.82 ± 0.25 | 1 | 0.91 | 0.96 | 8.1 | |
| Ext | 1.98± 0.24 | 1.83 ± 0.24 | 1 | 0.92 | 0.94 | 8 | 1.98 ± 0.24 | 1.83 ± 0.24 | 1 | 0.92 | 0.94 | 8 | ||
| 9 | Flex | 2.07± 0.25 | 1.92 ± 0.26 | 0.92 | 0.93 | 0.9 | 9.6 | 2.07 ± 0.25 | 1.92 ± 0.26 | 0.92 | 0.93 | 0.9 | 9.6 | |
| Ext | 2.12± 0.26 | 1.92 ± 0.26 | 1 | 0.91 | 0.91 | 9.2 | 2.12 ± 0.26 | 1.92 ± 0.26 | 1 | 0.91 | 0.91 | 9.2 | ||
Table II shows the difference of acceleration during activity between on-land and aquatic environments at different contraction phases. There is a significantly lower acceleration (7.9-9.6%) of the left and right arm in the aquatic environments. In addition, acceleration is observed during the starting (flexion) and ending (extension) of the onset duration as depicted in In Fig. 3. Hence, as it is shown in Table II, there are two occurrences of acceleration for every set of activities. There are no significant differences in the acceleration of two arms during the exercises in any environment. However, ICC and CV% ranged between r = 0.9–0.97 and 7.9–9.2%, respectively, between environments for the two protocols.
IV. DISCUSSION
The result of this study demonstrates that the environments affect the activities of BB muscles in terms of acceleration and onset duration. It is obvious that the acceleration of the arm movements is lower in the aquatic environments than in on land, and the onset duration of the activities decreases in the aquatic environments accordingly. As shown in Table I and Table II, the F-ratio is closer to 1 in most of the times which means that the variation among aquatic and on-land data comes from the same variance at the 5% significance level. The influence of drag force in terms of water resistance is one potential explanation for reduced muscle activity in water [5], [14]. In aquatic exercise, a faster movement velocity would result in a higher drag force, which would increase the movement’s resistance [14]. In our study, the participants completed one rep in 3-5 seconds based on the protocol as it is presented in In Fig. 3. It is found that the comparatively modest arm movement velocity would be able to lead to a smaller drag force and, consequently, less resistance to the squatting activities. The drag force might not be large enough to generate adequate resistance to enhance muscular activity since other aquatic factors, such as buoyancy, also changed the movement resistance [14]. The sEMG activity of the muscles in aquatic environment at various speeds could be the subject of future research.
To assess sEMG onset duration and acceleration reproducibility and precision for BB muscle, ICC, and intra-subject CV% were obtained [15]. For onset duration in Table I, the reproducibility and precision of the sEMG recordings for BB muscle were observed high (ICC= 0.93-0.98 and CV%= 2.7-6.4 %). These results are consistent with those of other studies carried out comparable reproducibility tests on-land and in aquatic environment [2]. In addition, to observe the reproducibility and precision of the acceleration of arm movement during sEMG of BB, the findings were high (ICC= 0.9-0.97 and CV%= 7.9-9.2 %) as well.
It’s noteworthy for clinicians or therapists to know that performing an exercise in an aquatic environment could reduce limb muscle activities when designing a rehabilitation program for patients with joint injuries who present with a high pain score or limb weakness, particularly in the early stages of the injury or during the postoperative period. By understanding that exercising speed and immersion depth affects how much muscle activity is reduced, further research in this area may aid in adjusting these variables to create the best rehabilitation plan for persons with limb difficulties [16], [17]. Regarding limitations, as only 5 males participated in the experiments, this is a much smaller sample to reach a statistical decision. Only two experimental protocols and one muscle are not enough to reach a final decision regarding the adoption of the developed device for musculoskeletal monitoring during aquatic therapy and rehabilitation.
V. CONCLUSION
Based on the described methodologies and device architecture, comparable MVC sEMG values are attainable on-land and in aquatic environment using developed device. Moreover, onset duration and acceleration values are very reproducible across land and aquatic testing conditions. The outcomes show the suitability of the developed device for musculoskeletal monitoring during aquatic therapy and rehabilitation. Due to the potential impact of activities of different muscles in signal integrity, we would further proceed with MVC sEMG studies for different muscles in land and aquatic environments. Angular movement monitoring is another subject of monitoring to verify the correctness of the activities during aquatic therapy and rehabilitation.
Fig. 1.
Wearable sEMG and IMU recording device with (a) Flash memory part on the right wing and switch to control the data collection on the left wing and wirelessly rechargeable battery on middle (Top view), (b) CB/PDMS electrodes on two wings connected with flexible part and wireless charging coil (Bottom view), (c) flexibility of the device (side view).
Fig. 2.
Experimental protocol with a device attached on both arms of a participant performing aquatic exercise (a) Extension and (b) flexion in protocol 1 (c) extension and (d) flexion in protocol 2.
Acknowledgments
This work was funded by NIH grant #P2CHD101899, which is supported by Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) and National Institute of Neurological Disorders And Stroke (NINDS).
References
- [1].Kaneda K, Ohgi Y, Mckean M, and Burkett B, “Underwater electromyogram for human health exercise,” in Electrodiagnosis in New Frontiers of Clinical Research. IntechOpen, 2013. [Google Scholar]
- [2].Silvers WM and Dolny DG, “Comparison and reproducibility of semg during manual muscle testing on land and in water,” Journal of electromyography and kinesiology, vol. 21, no. 1, pp. 95–101, 2011. [DOI] [PubMed] [Google Scholar]
- [3].Iliescu AM, McIntyre A, Wiener J, Iruthayarajah J, Lee A, Caughlin S, and Teasell R, “Evaluating the effectiveness of aquatic therapy on mobility, balance, and level of functional independence in stroke rehabilitation: a systematic review and meta-analysis,” Clinical rehabilitation, vol. 34, no. 1, pp. 56–68, 2020. [DOI] [PubMed] [Google Scholar]
- [4].Xue J, Yang Y, Chen J, Jiang Y, Zhu C, Yokoi H, and Duan F, “The development of an underwater semg signal recognition system based on conductive silicon,” in 2019 IEEE International Conference on Advanced Robotics and its Social Impacts (ARSO). IEEE, 2019, pp. 387–392. [Google Scholar]
- [5].Yuen CH, Lam CP, Tong KC, Yeung JC, Yip CH, and So BC, “Investigation the emg activities of lower limb muscles when doing squatting exercise in water and on land,” International journal of environmental research and public health, vol. 16, no. 22, p. 4562, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Silva MF, Dias JM, Bela LFD, Pelegrinelli ARM, Lima TB, da Silva Carvalho RG, Taglietti M, Júnior JPB, Facci LM, McVeigh JG et al. , “A review on muscle activation behaviour during gait in shallow water and deep-water running and surface electromyography procedures,” Journal of Bodywork and Movement Therapies, vol. 24, no. 4, pp. 432–441, 2020. [DOI] [PubMed] [Google Scholar]
- [7].Cuesta-Vargas A, Cano-Herrera C, and Heywood S, “Analysis of the neuromuscular activity during rising from a chair in water and on dry land,” Journal of Electromyography and Kinesiology, vol. 23, no. 6, pp. 1446–1450, 2013. [DOI] [PubMed] [Google Scholar]
- [8].Posada-Quintero H, Noh Y, Eaton-Robb C, Florian JP, and Chon KH, “Feasibility testing of hydrophobic carbon electrodes for acquisition of underwater surface electromyography data,” Annals of biomedical engineering, vol. 46, pp. 1397–1405, 2018. [DOI] [PubMed] [Google Scholar]
- [9].Duan F, Xue J, Sun Z, Chen X, Liu Y, and Yokoi H, “A novel waterproof semg electrode based on conductive silicone for underwater signal recognition,” IEEE Sensors Journal, vol. 21, no. 18, pp. 20 388–20 397, 2021. [Google Scholar]
- [10].Psycharakis SG, Coleman SG, Linton L, Kaliarntas K, and Valentin S, “Muscle activity during aquatic and land exercises in people with and without low back pain,” Physical Therapy, vol. 99, no. 3, pp. 297–310, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Kaneda K, Wakabayashi H, Sato D, and Nomura T, “Lower extremity muscle activity during different types and speeds of underwater movement,” Journal of physiological anthropology, vol. 26, no. 2, pp. 197–200, 2007. [DOI] [PubMed] [Google Scholar]
- [12].Noh YS, Cho CH, Salehizadeh S, Reyes BA, Bales JR, and Chon KH, “A novel cb/pdms electrode for ecg monitoring during swimming: 399 board# 236 june 1, 11: 00 am-12: 30 pm,” Medicine & Science in Sports & Exercise, vol. 48, no. 5S, p. 109, 2016. [Google Scholar]
- [13].Reyes BA, Posada-Quintero HF, Bales JR, Clement AL, Pins GD, Swiston A, Riistama J, Florian JP, Shykoff B, Qin M, and Chon KH, “Novel electrodes for underwater ecg monitoring,” IEEE Transactions on Biomedical Engineering, vol. 61, no. 6, pp. 1863–1876, 2014. [DOI] [PubMed] [Google Scholar]
- [14].Alberton CL, Cadore EL, Pinto SS, Tartaruga MP, Da Silva EM, and Kruel LFM, “Cardiorespiratory, neuromuscular and kinematic responses to stationary running performed in water and on dry land,” European Journal of Applied Physiology, vol. 111, pp. 1157–1166, 2011. [DOI] [PubMed] [Google Scholar]
- [15].Norcross MF, Blackburn JT, and Goerger BM, “Reliability and interpretation of single leg stance and maximum voluntary isometric contraction methods of electromyography normalization,” Journal of electromyography and kinesiology, vol. 20, no. 3, pp. 420–425, 2010. [DOI] [PubMed] [Google Scholar]
- [16].Roca-Dols A, Losa-Iglesias ME, Sánchez-Gómez R, Becerro-de Bengoa-Vallejo R, López-López D, Palomo-López P, Rodríguez-Sanz D, and Calvo-Lobo C, “Electromyography activity of triceps surae and tibialis anterior muscles related to various sports shoes,” Journal of the Mechanical Behavior of Biomedical Materials, vol. 86, pp. 158–171, 2018. [DOI] [PubMed] [Google Scholar]
- [17].Roca-Dols A, Losa-Iglesias ME, Sánchez-Gómez R, López-López D, Becerro-de Bengoa-Vallejo R, and Calvo-Lobo C, “Electromyography comparison of the effects of various footwear in the activity patterns of the peroneus longus and brevis muscles,” Journal of the Mechanical Behavior of Biomedical Materials, vol. 82, pp. 126–132, 2018. [DOI] [PubMed] [Google Scholar]