Simulation of Lower Limb Muscle Activity During Inclined Slope Walking

(1)

IN

DEGREE PROJECT MEDICAL ENGINEERING, SECOND CYCLE, 30 CREDITS

,

STOCKHOLM SWEDEN 2019

Simulation of lower limb muscle

activity during inclined slope

walking

GANESH PRASANTH ARUMUGANAINAR

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ENGINEERING SCIENCES IN CHEMISTRY, BIOTECHNOLOGY AND HEALTH

(2)

(3)

Simulation of lower limb muscle

activity during inclined slope

walking

GANESH PRASANTH ARUMUGANAINAR

Degree Project in Medical Engineering Stockholm, Sweden 2018

External Supervisor: Elena Gutierrez-Farewik Group Supervisor: Rodrigo Moreno

Examiner: Mats Nilsson Reviewer: Svein Kleiven

School of Engineering Sciences in Chemistry, Biotechnology and Health KTH Royal Institute of Technology

(4)

(5)

Abstract

Robotic exoskeletons are designed to assist patients with motor dysfunctions. Recent researches focus on extending the robotic assistance to patient activities other than ground level walking. This study aims to analyse the lower limb muscle activity during inclined slope walking contrasting with that of ground level walking. Two different angles of inclination were chosen: 9 degrees and 18 degrees. 9 degrees inclined slope is the universal ramp size for wheelchairs. The hypothesis is that muscle activation, and ultimately metabolic cost, in inclined slope walking is different from that of ground level walking. Collected motion data and simulation in OpenSim prove that the difference in metabolic cost is because of increased activity of ankle dorsiflexors and hip extensors and reduced activity of knee extensors. Finally, muscle activities along with other criteria such as kinematic alignment and joint range of motion are summed up as biomechanical considerations for robotic exoskeleton design.

Sammanfattning

Robotiska exoskeletoner är utformade för att hjälpa patienter med motorisk dysfunktion. Nyare undersökningar fokuserar på att utöka robotassistansen till andra patientaktiviteter än grundnivåvandring. Denna studie syftar till att analysera muskelaktiviteten för nedre extremiteten under gång i lutning jämfört med gång på plan yta. Två olika lutningsvinklar valdes: 9 grader och 18 grader. 9 graders lutning är den universella rampstorleken för rullstolar. Hypotesen är att muskelaktivering och slutligen metabolisk kostnad, under gång i lutning, skiljer sig från gång på plan yta. Insamlad rörelsedata och simulering i OpenSim bevisar att skillnaden i metabolisk kostnad är på grund av ökad aktivitet av fotled dorsiflexorer och höft extensorer och minskad aktivitet av knäextensorer. Slutligen summeras muskelaktiviteter tillsammans med andra kriterier som kinematisk inriktning och gemensamt rörelseområde som biomekaniska överväganden för robotisk exoskelettdesign.

(6)

(7)

Acknowledgements

I wholeheartedly thank my supervisor Elena Gutierrez-Farewik for giving me an opportunity to work on this project at KTH Mechanics. I am always grateful for the passion she showed towards the project and the friendly ambience she created in the group.

I thank my parents, family and friends for their immense love and constant support though miles away. A special thanks to my friends who willingly participated as test subjects in my experiment.

My sincere thanks to Mikael Remeringen and Cecilia Lidbeck at the Motoriklab, Astrid Lindgren’s Barnsjukhus for their valuable guidance and precious time spent in providing the laboratory support.

My wholehearted thanks to Rodrigo Moreno at School of Technology and Health, KTH for his constructive feedback during group supervision sessions. A special thanks to my thesis reviewer Svein Kleiven.

Annie Charles and Katya Mehyeddine earn a special acknowledgement for their valuable time spent in translating my abstract into Swedish.

(8)

(9)

2 Materials and Methods 1 2.1 Subject . . . 1 2.2 Experimental Setup . . . 2 2.3 Testing Protocol . . . 2 2.4 Data Processing . . . 3 3 Results 5 3.1 Considerations at Ankle . . . 5 3.1.1 Degrees of Freedom . . . 6 3.1.2 Range of Motion . . . 6 3.1.3 Muscle Activity . . . 6 3.2 Considerations at Knee . . . 8 3.2.1 Degrees of Freedom . . . 8 3.2.2 Range of Motion . . . 9 3.2.3 Muscle Activity . . . 9 3.3 Considerations at Hip . . . 9 3.3.1 Degrees of Freedom . . . 9 3.3.2 Range of Motion . . . 10 3.3.3 Muscle Activity . . . 10 3.4 Validation of Results . . . 11 4 Discussion 12

5 Limitations and Future Work 13

6 Conclusion 14

Bibliography 14

Appendices 17

(10)

A Literature Study 17

A.1 Introduction . . . 17

A.2 Inclined Slope Walking . . . 17

A.2.1 Phases . . . 17

A.2.2 Biomechanical Differences . . . 17

A.3 Tools . . . 19

A.3.1 Marker placement . . . 19

A.3.2 Software Tools . . . 19

A.4 Current trends in robotic exoskeletons . . . 20

A.4.1 Gait rehabilitation . . . 20

A.4.2 Human locomotion assistance . . . 21

A.4.3 Human strength augmentation . . . 21

A.5 Limitations and challenges . . . 22

A.6 Summary . . . 23

A.7 References . . . 23

B Muscle Activity Plots 25

(11)

List of Figures

2.1 Wooden ramps placed on the force plates . . . 2 2.2 Sequence of events in one gait cycle . . . 3 2.3 A screenshot of Lee-Son’s Toolbox interface . . . 4 3.1 Kinematics at ankle for both trials during ground level walking

and inclined slope walking (Positive values denote dorsiflexion levels and negative values represent plantarflexion levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.) . . . 5 3.2 Comparison of ankle dorsiflexion and plantarflexion peaks during

ground level and inclined slope walking (positive values denote dorsiflexion levels and negative values represent plantarflexion levels) . . . 6 3.3 Comparison of Soleus muscle activity peaks during ground level

and inclined slope walking (The peak values of both the trials and their mean are given) . . . 7 3.4 Comparison of Tibialis Posterior muscle activity peaks during

ground level and inclined slope walking (The peak values of both the trials and their mean are given) . . . 7 3.6 Comparison of knee flexion and extension peaks during ground

level and inclined slope walking (positive values denote extension levels and negative values represent flexion levels) . . . 8 3.5 Kinematics at knee for both trials during ground level walking

and inclined slope walking (Positive values denote extension levels and negative values represent flexion levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.) . . . 8 3.7 Comparison of Rectus Femoris muscle activity peaks during

ground level and inclined slope walking (The peak values of both the trials and their mean are given) . . . 9 3.8 Kinematics at hip for both trials during ground level walking and

inclined slope walking (Positive values denote flexion levels and negative values represent extension levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.) . . . 10 3.10 Comparison of Adductor Magnus 2 muscle activity peaks during

ground level and inclined slope walking (The peak values of both the trials and their mean are given) . . . 11 3.9 Comparison of hip flexion and extension peaks during ground

level and inclined slope walking (positive values denote flexion levels and negative values represent extension levels) . . . 11

(12)

3.11 Comparison of EMG peak values of Rectus Femoris, Biceps

Femoris and Tibialis Anterior . . . 12

A.1 Markers location (Markers used in the study are framed in red) . 20 A.2 Table summarizing currently available lower extremity exoskeletons 22 B.1 Soleus muscle activation comparison (Trial 1) . . . 25

B.2 Soleus muscle activation comparison (Trial 2) . . . 25

B.3 Tibialis Posterior muscle activation comparison (Trial 1) . . . 26

B.4 Tibialis Posterior muscle activation comparison (Trial 1) . . . 26

B.5 Rectus Femoris muscle activation comparison (Trial 1) . . . 27

B.6 Rectus Femoris muscle activation comparison (Trial 2) . . . 27

B.7 Adductor Magnus muscle activation comparison (Trial 1) . . . . 28

B.8 Adductor Magnus muscle activation comparison (Trial 1) . . . . 28

(13)

List of Tables

2.1 Anthropometric data of the chosen subject as on the day of recording . . . 1 A.1 Markers Description . . . 20

(14)

List of Abbreviations

EMG - Electromyogram SCI - Spinal Cord Injury CSV - Comma Separated Values CMC - Computed Muscle Control

(15)

HL205X Degree Project in Medical Engineering

1 Introduction

Improper gaits are a result of stroke, accidents, Parkinson’s, arthritis or multiple sclerosis[1]. There is a wide possibility for correcting these improper gait patterns through rehabilitation. Among various methods, efficient robotic exoskeleton assistance stands tall. Studies are done to improve the communication interface between the user and the robotic exoskeleton device[5]. Majority of these studies focus on normal gaits but not on other complex motions such as walking up inclined slopes. It is important to extend the robotic assistance to inclined slope walking rather than just ground level walking. This particular study aims to analyse lower limb muscle activity during walking up inclined slopes and contrast the difference with that of ground level walking.

1.1 Aim

The aim of this study is to analyse how lower limb muscle activity differs between walking on an inclined slope and ground level walking.

1.2 Hypothesis

The hypothesis is that muscle activity, and ultimately metabolic cost, in inclined slope walking is different from that of ground level walking, and the difference in metabolic cost is because of increased activity of ankle plantarflexors and hip extensors and reduced activity of knee extensors (It is enough to consider one from each group hip flexors/knee extensors and hip extensors/knee flexors because the muscles responsible for hip extension are also responsible for knee flexion and similarly the muscles responsible for hip flexion are also responsible for knee extension).

2 Materials and Methods

2.1 Subject

Table 2.1: Anthropometric data of the chosen subject as on the day of recording

Attribute Value Age (years) 28 Sex (M/F) M Height (cm) 175 Weight (kg) 73 Body Mass Index (kg/m2₎ _23.8

A subject who is healthy, active and having no history of lower extremity injury was selected from the students of KTH Royal Institute of Technology.

(16)

Anthropometric data of the selected subject were measured on the day of recording the motion. The data are tabulated in table 2.1.

2.2 Experimental Setup

Motion data are captured in the ’Motorik Lab’ of Astrid Lindgren’s Barnsjukhus, Solna. The laboratory has an approximate 15 meter walkway with two Kistler force plates sampled at 1000 Hz frequency in the middle of the walkway. Kinematic data of the performed motions are collected by a 8 camera VICON Motion Capture System coupled with VICON Nexus software. For inclined slope walking, wooden ramps of two different angles of inclination: 9 degrees and 18 degrees were constructed. They were placed on force plates as shown in figures 2.1(a) and 2.1(b). Force plates are neutralised after placing the ramps so that weights of the wooden pieces are eliminated during the calculation of external loads.

Electromyogram (EMG) data are also recorded with an aim of validating the results obtained from OpenSim later. The EMG data are acquired through Noraxon DTS EMG sensor system, which is coupled with VICON Nexus software. Three different lower limb muscles are chosen for EMG: Rectus Femoris (a hip flexor and knee extensor), Biceps Femoris (a hip extensor and knee flexor) and Tibialis Anterior (an ankle dorsiflexor). The EMG sensors are placed on the skin as per the directions given by Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscles (SENIAM)[21].

(a) 9 degree inclination (b) 18 degree inclination Figure 2.1: Wooden ramps placed on the force plates

2.3 Testing Protocol

The test subject was prepared with 23 markers placed on trunk and lower limbs as per Plug-in Gait model requirements[10] and three EMG sensors on the three different muscles as mentioned earlier. Table A.1 in the appendix represents the 23 markers used and figure A.1 shows the location of those markers. The subject walked several times on the walkway and the ramps to

(17)

get comfortable with.

The subject with placed markers and EMG sensors was allowed to perform three different motions: ground level walking, walking up 9 degrees inclined slope and walking up 18 degrees inclined slope. The subject was made to place the right foot on the first force plate when walking. Seven trials were recorded for each motion. Recorded trials were discarded if the foot was not fully placed on the plates or the subject forcefully altered the walking pattern in order to place the right foot on the right plate. Two successful trials for the chosen subject are considered for analysis. One gait cycle is considered as the sequence of events from first right toe strike to second right toe strike as shown in figure 2.2[23]. So as per the considerations, analysis are done for the sequence of events from first right toe strike on first force plate, left toe strike on second force plate and second right toe strike on the walkway. For a healthy individual, gait cycle is symmetrical about the sagittal plane and hence considering only the right gait cycle for analysis proves to be sufficient[24].

Figure 2.2: Sequence of events in one gait cycle

2.4 Data Processing

VICON Nexus collects motion data of the 23 markers and the EMG data from the 3 sensors placed on the subject. The collected trial needs to be reconstructed first. There is a higher probability that the marker information might go missing in some time frames during the trial due to the quality of the camera system or any interference that blocked the camera from capturing the marker data completely. If the trials are exported with unlabelled markers and unfilled gaps, OpenSim would not be able to simulate the desired kinematics and kinetics. For these reasons, the quality of the collected trial needs reviewing and processing prior to be exported from VICON Nexus. The collected trial is reviewed to find if there are any unlabelled markers and missing trajectories of the markers. The amount of processing required depends upon the quality of the captured data[15]. Automatic labelling tool labels all the markers. Still one or more marker might go missing and needs manual labelling. Once the markers are labelled, gaps in the motion path of each marker are to be identified and filled. There are both automatic and manual options for gap filling the trajectories. Auto-gap fill option works best for smaller gaps. It is advisable to use manual gap filling for larger gaps[16]. There are five different gap filling tools available for manual gap filling: Spline fill, pattern fill, Rigid body fill, Kinematic fill, Cyclic fill. Suitable gap filling

(18)

tool that works best for the particular case is chosen as per the guidelines from VICON Nexus User Guide[16]. Once the trials of our interest are completely labelled and gaps filled, they are exported from VICON Nexus as Comma Separated Values ’.CSV’ file.

This ’.CSV’ file contains the whole trajectory information of the 23 markers, force plate data and EMG data. This ’.CSV’ file needs to be converted to OpenSim usable file formats. Lee-Son’s Toolbox is an open-source toolkit that adapts to the number of markers, force plates and the global coordinates and easily convert them to ’.trc’ file containing marker trajectories and ’.mot’ file containing force plate data which can be used directly in OpenSim. A screenshot of the Lee-Son Toolbox interface is shown in the figure 2.3. This is a user-friendly toolkit that adapts to OpenSim and laboratory coordinate axes and the number of markers and convert ’.CSV’ files into OpenSim usable file formats[18].

Figure 2.3: A screenshot of Lee-Son’s Toolbox interface

OpenSim is an open-source musculoskeletal modelling platform[14]. The various tools available in OpenSim and what they are meant for are explained in the appendix A.3.2. Computed Muscle Control (CMC) tool in OpenSim computes muscle excitation levels necessary to drive the model to perform the desired kinematics as per the marker data in the presence of external loads as per the force plate data.

A brief explanation of how CMC tool algorithm works is given below. The first step is to compute desired accelerations necessary to drive the musculoskeletal model to perform the desired kinematics. Since, there cannot be an instantaneous change in the muscle forces applied to the body, desired accelerations are computed for a specific time interval (usually 0.01 seconds) which is long enough for the muscle forces to change. The next step in CMC is to compute the actuator controls that achieve the previously computed desired accelerations. The final step of CMC is to conduct a forward dynamic

(19)

simulation advancing forward in time. The CMC algorithm is iterative where these three steps are repeated until the end of desired movement.

Output from CMC execution has the computed kinematics at joints and activity of individual muscles during the desired motions which are then plotted and compared in Matlab. While plotting, only the peak values matter more than the activity pattern so as to decide the activity levels of muscles. Moreover, the designers of robotic exoskeleton too consider the peaks for sizing their actuators[22].

3 Results

Ground level walking differs from inclined slope walking in terms of kinematics at hip, knee and ankle. The designers of robotic exoskeleton would be interested in degrees of freedom, range of motion and muscular activity associated with each joint in order to design a robotic exoskeleton that complies with human limb.

3.1 Considerations at Ankle

(a) Trial 1 (b) Trial 2

Figure 3.1: Kinematics at ankle for both trials during ground level walking and inclined slope walking (Positive values denote dorsiflexion levels and negative values represent plantarflexion levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.)

(20)

Figure 3.2: Comparison of ankle dorsiflexion and plantarflexion peaks during ground level and inclined slope walking (positive values denote dorsiflexion levels and negative values represent plantarflexion levels)

3.1.1 Degrees of Freedom

Ankle is a hinge joint that has only one degree of freedom- dorsiflexion and plantarflexion.

3.1.2 Range of Motion

Figure 3.1 shows the ankle angle variation over percent gait cycle for the two trials considered (positive values denote dorsiflexion levels and negative values represent plantarflexion levels). Also, it is observed that the range of motion for ankle goes from -14.748 to 9.278 degrees for ground level, from -15.510 to 15.342 degrees for walking up 9 degree slope and from -15.045 to 34.37 degrees for walking up 18 degree slope. Also, the dorsiflexion and plantarflexion levels are greater for walking up 18 degree inclined slope. The dorsiflexion level increases with increasing slope with walking up 18 degree slope being the highest and ground level walking being the lowest. Whereas, the plantarflexion level is almost the same for ground level walking and walking up 9 degree inclined slope. The ankle plantarflexes more during walking up 18 degree inclined slope.

3.1.3 Muscle Activity

The muscle activity associated could be analysed by considering two different muscles responsible for ankle plantarflexion: Soleus (figure 3.3) and Tibialis Posterior (figure 3.4). In figures 3.3 and 3.4, it is clearly observed that the muscle activity peak is the highest for walking up 18 degree slope and subsequently decreases with decrease in slope which is in accordance with the kinematic results shown in figures 3.1 and 3.2. The smaller standard deviation of Soleus muscle activity during walking up 18 degree inclined slope could be seen as a

(21)

result of the subject recreating the same level of ankle movement in both the trials.

Figure 3.3: Comparison of Soleus muscle activity peaks during ground level and inclined slope walking (The peak values of both the trials and their mean are given)

Figure 3.4: Comparison of Tibialis Posterior muscle activity peaks during ground level and inclined slope walking (The peak values of both the trials and their mean are given)

(22)

Figure 3.6: Comparison of knee flexion and extension peaks during ground level and inclined slope walking (positive values denote extension levels and negative values represent flexion levels)

3.2 Considerations at Knee

Figure 3.5: Kinematics at knee for both trials during ground level walking and inclined slope walking (Positive values denote extension levels and negative values represent flexion levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.)

Knee is a condyloid joint and has two rotational degrees of freedom-flexion/extension and internal/external rotation. However, previous literature consider only flexion/extension as the only one degree of freedom due to reduced internal/external rotations at knee.

(23)

Figure 3.5 shows the knee angle variation over percent gait cycle for the two trials considered (positive values denote extension levels and negative values represent flexion levels). Also, it is observed that the range of motion for knee goes from -52.413 to 1.546 degrees for ground level, from -58.007 to -2.454 degrees for walking up 9 degree slope and from -79.695 to -5.143 degrees for walking up 18 degree slope. In figure 3.5, it is observed that the knee flexion is the greatest for walking up 18 degree inclined slope and subsequently decreases with decease in slope. In figure 3.6, it is observed that the knee extension is the highest during ground level walking and it decreases with increase in slope. This is because the presence of wooden ramp restricts knee from fully extending, thereby reducing the knee extension angle.

The muscle activity associated could be analysed by considering Rectus Femoris, a knee extensor muscle. In figure 3.7, it is evident that the muscle activity is the highest for ground level walking and it decreases with increase in slope which is in accordance with the kinematic results shown in figures 3.5, 3.6. Knee extension activity is reduced for increasing slopes of inclination (figures 3.5 and 3.6) and so the muscle activity associated with it.

Figure 3.7: Comparison of Rectus Femoris muscle activity peaks during ground level and inclined slope walking (The peak values of both the trials and their mean are given)

3.3 Considerations at Hip

Hip is a ball and socket joint that has three degrees of freedom-flexion/extension, adduction/abduction and internal/external rotation.

(24)

However, for convenience only flexion/extension of hip are analysed in detail and other two degrees of freedom bring scope for future study.

Figure 3.8 shows the hip angle variation over percent gait cycle for the two trials considered (positive values denote flexion levels and negative values represent extension levels). Also, it is observed that the range of motion for hip goes from -23.156 to 16.303 degrees for ground level, from -14.338 to 37.904 for walking up 9 degree slope and from -8.26 to 60.94 for walking up 18 degree slope. Figure 3.8 shows that hip flexion is the lowest for ground level walking and it increases with increase in slope. Also, that hip extension is the highest for ground level walking and it decreases with increase in slope.

The muscle activity associated with this could be analysed by considering Adductor Magnus 2, a hip extensor. Recent studies focus on Adductor Magnus extending its function for activities other than adduction. This gained attraction and that’s the reason why Adductor Magnus is chosen over other hip muscles for analysis. The activity peak of Adductor Magnus is almost the same for ground level walking and inclined slope walking when observed visually. However, their comparison is explained statistically in the ’Discussion’ section.

Figure 3.8: Kinematics at hip for both trials during ground level walking and inclined slope walking (Positive values denote flexion levels and negative values represent extension levels. First right toe strike at 0 % and second right toe strike at 100% gait, Mid-stance at 23% gait and right heel off at 45% gait.)

(25)

Figure 3.10: Comparison of Adductor Magnus 2 muscle activity peaks during ground level and inclined slope walking (The peak values of both the trials and their mean are given)

Figure 3.9: Comparison of hip flexion and extension peaks during ground level and inclined slope walking (positive values denote flexion levels and negative values represent extension levels)

3.4 Validation of Results

In order to validate the results obtained from OpenSim, EMG is acquired from three different muscles: Rectus Femoris, Biceps Femoris and Tibialis Anterior to confirm the activity levels. The EMG results coincided with the OpenSim results. A graph comparing the mean of peak EMG values of the above mentioned three muscles during ground level walking and inclined slope walking for two trials is shown in figure 3.11. It is evident that the activity of Biceps Femoris, a hip extensor and Tibialis Anterior, an ankle dorsiflexor has

(26)

increased with increase in slope whereas the activity of Rectus Femoris, a knee extensor has decreased with increase in slope.

Figure 3.11: Comparison of EMG peak values of Rectus Femoris, Biceps Femoris and Tibialis Anterior

4 Discussion

Considering the activity of Soleus, there is a 3.57% increase in the activity level during walking up 9 degree inclined slope compared to that of ground level walking. Whereas, there is a whopping 78.5% increase in activity during walking up 18 degree inclined slope compared to ground level walking. This increase in muscle activity levels are statistically significant with (p<0.45) and (p<0.11) respectively.

Considering the activity of Tibialis Posterior, there is a 48.5% increase in activity during walking up 9 degree inclined slope compared to ground level walking. There is a 54.2% increase in activity during walking up 18 degree inclined slope compared to that of ground level walking. This increase in muscle activity levels are found to be statistically significant with (p<0.13) and (p<0.11) respectively. Soleus and Tibialis Posterior are ankle plantarflexors and this increase in activity level is in accordance with the ankle plantarflexion angles in figure 3.1. This could be due to the increased ankle dorsiflexion levels during inclined slope walking due to the slope of ramps. This in turn causes the plantarflexors to activate more for plantarflexing the ankle.

Considering the activity of Rectus Femoris, there is a 7.14% decrease during walking up 9 degree slope and a 10% decrease during walking up 18 degree inclined slope when compared to ground level walking. This decrease in muscle activity levels are found to be statistically significant with (p<0.42) and (p<0.39) respectively. The larger values of alpha in this case may be due to

(27)

the consideration of less number of samples and it might change when more number of trials are considered. The slope of the ramp restricts the knee from extending completely and hence the activity of the knee extensor decreases with increase in slope. This result is in accordance with the EMG results shown in figure 3.11 and kinematic results shown in figure 3.6.

Considering the activity of Adductor Magnus 2, it is difficult to generalise the difference in activity with respect to slope since the there is no much difference seen visually. However, the muscle activity during walking up 9 degree slope is greater than that of ground level walking with statistical significance (p<0.46). Also, the muscle activity during walking up 18 degree slope is greater than that of ground level walking with statistical significance (p<0.54). With greater significance values, the results may not be generalised at this point but in future with more trials or subjects considered, this activity difference could be seen prominent and that would help to conclude how the muscle activity at hip varies with respect to slope.

With these simulated OpenSim results, validation through acquired EMG data and statistical hypothesis testing, proposed hypotheses are verified. The difference in metabolic cost between ground level walking and inclined slope walking is due to the fact that the ankle plantarflexors have an increased muscle activity and knee extensors have a decreased muscle activity during inclined slope walking compared to ground level walking. For muscle activity at the hip, the results are not convincing enough to generalise the hypothesis at this point.

5 Limitations and Future Work

One of the biggest limitation of the work is that the study is performed on only one healthy subject considering only two valid trials. Had more trials been considered, the results would have become more reliable. Another limitation is that the statistical significance proving the difference in muscle activity levels is higher comparing to the usual 0.05. This might also be an effect of considering less number of trials. Had more number of trials been considered, the difference in muscle activity levels would be more prominent thereby giving a better statistical significance level. The overall idea is to provide information to develop a robotic exoskeleton that assists patients with motor dysfunctions in performing complex motions like climbing up an inclined slope. So, there are also biomechanical considerations other than degrees of freedom, range of motion and muscle activity at individual joints to be considered such as joint torque, its application at the right time with right intensity and the velocity with which the joint angles change. Also, it is necessary to study how patients with muscle weaknesses and other abnormalities adapt to similar kind of motion. Finally, it is necessary to study other muscle activities too in order to analyse the combined activity at

(28)

individual joints. When these factors are considered and analysed, the robotic exoskeleton shall be designed in such a way that it provides torque to the limb joints whenever necessary thereby assisting patients with motor dysfunctions in climbing up inclined slopes.

6 Conclusion

This study is aimed to analyse the difference in lower limb muscle activity during ground level walking and walking up an inclined slope. Simulation results from OpenSim software clearly show that muscle activity of ankle plantarflexors has increased with increase in slopes whereas the activity of knee extensors has decreased with increase in slopes. These results are in accordance with the acquired EMG data from three different muscles and also found to be statistically significant. And this verifies the proposed hypothesis. Further research is necessary in analysing other factors such as the combined muscle activity at a joint, joint torque and velocity with which the joint angle changes thereby predicting the requirements for a robotic exoskeleton to assist patients with motor dysfunctions in performing the desired complex motions such as walking up inclined slopes.

Bibliography

[1] Patricia Krawetz, MD, Patricia Nance, MD. Gait Analysis of Spinal Cord Injured Subjects: Effects of Injury Level and Spasticity. Arch Phys Med Rehabil Vol 77, July 1996.

[2] Amanda Ortiz. The metabolic cost of walking and running up a 30 degree incline: implications for vertical kilometer foot races. Undergraduate Honors Theses. 2017

[3] Maxim N. Nikolenko, Denis A. Kotin. General Principles of Medical Exoskeleton Design. 18th International Conference on Micro/Nanotechnologies and Electrical Devices. 2017.

[4] Ian Benson, Kirsten Hart1, Dot Tussler and Joost J van Middendorp. Lower-limb exoskeletons for individuals with chronic spinal cord injury: findings from a feasibility study. Clinical Rehabilitation 2016, Vol. 30(1) 73–84.

[5] Chunjie Chen, Xinyu Wu, Du-xin Liu, Wei Feng and Can Wang. Design and Voluntary Motion Intention Estimation of a Novel Wearable Full-Body Flexible Exoskeleton Robot. Hindawi Mobile Information Systems Volume 2017.

(29)

[6] Sadiq J. Abbas and Zahraa M. Abdulhassan. Kinematic Analysis of Human Climbing up and Down Stairs at Different Inclinations. Eng. Tech.Journal, Vol. 31,Part (A), No.8, 2013.

[7] Emilie Desrosiers, Cyril Duclos and Sylvie Nadeau. Gait adaptation during walking on an inclined pathway following spinal cord injury. Clinical Biomechanics 29 (2014) 500–505.

[8] Bing Chen, Hao Ma, Lai-Yin Qin, Fei Gao, Kai-Ming Chan,Sheung-Wai Law, Ling Qin, Wei-Hsin Liao. Recent developments and challenges of lower extremity exoskeletons. Journal of Orthopaedic Translation (2016) 5, 26-37.

[9] Kotaro Sasaki, Richard R Neptune. Muscle mechanical work and elastic energy utilization during walking and running near the preferred gait transition speed. Gait Posture 23 (2006) 383–390.

[10] Vicon Motion System. Modeling with Plug-in Gait. 2018.

URL:https://docs.vicon.com/display/Nexus25/Modeling+with+Plug- in+Gaithttps://docs.vicon.com/display/Nexus25/Modeling+with+Plug-in+Gait.

[11] National Center for Simulation in Rehabilitation Research. Gait 2392 and 2354 Models. URL: https://simtk-confluence.stanford.edu/display/OpenSim/Gait+2392+and+2354+Models.

[12] National Center for Simulation in Rehabilitation Research. OpenSim User’s Guide. 2017. URL: https://simtk-confluence.stanford.edu/display/OpenSim/User%27s+Guide

[13] Felipe Costa Alvim. c3d2OpenSim. 2014. URL: https://simtk.org/projects/c3d2opensim.

[14] Guide to OpenSim Workflow and Tools. 2018. URL:

https://simtk-confluence.stanford.edu/display/OpenSim/Guide+to+OpenSim+Workflow+and+Tools.

[15] Nexus 2.5 VICON Documentation Pipeline Tools. 2018. URL: https://docs.vicon.com/display/Nexus25/Pipeline+tools.

[16] VICON Nexus User Guide. 2016. URL: https://docs.vicon.com/display/Nexus25/Vicon+Nexus+User+Guide.

[17] Paul DeVita et al. Muscles do more positive than negative work in human locomotion. J Exp Biol. 2007 October ; 210(Pt 19): 3361–3373. doi:10.1242/jeb.003970.

(30)

[18] Lee Sang Yun, Jinkyou Son. Lee-Son’s Toolbox User Manual. URL: https://simtk.org/home/lee-son/.

[19] Tim Dorn. C3D Extraction Toolbox. URL: https://simtk.org/home/c3dtoolbox by Tim Dorn.

[20] Peter Loan. OpenSMAC: Utility for importing Motion Analysis data (TRB, ANB) into OpenSim. URL: https://simtk.org/home/opensmac.

[21] Seniam Group. SENIAM Project. http://www.seniam.org/.

[22] Massimo Cenciarini and Aaron M. Dollar. Biomechanical Considerations in the Design of Lower Limb Exoskeletons. IEEE International Conference on Rehabilitation Robotics. 2011.

[23] Walking in Graphs. University of Dallas. URL: https://www.utdallas.edu/atec/midori/Handouts/walkingGraphs.htm

[24] Kadek Heri Sanjaya et al. The biomechanics of walking symmetry during gait cycle in various walking condition. Proceedings of 2016 1st International Conference on Biomedical Engineering: Empowering Biomedical Technology for Better Future, IBIOMED 2016, , art. no. 7869813.

(31)

Appendices

A

Literature Study

A.1 Introduction

The report is an overall idea gained from the literature survey performed to understand how ground level walking differs from inclined slope walking in terms of lower limb muscle activity to predict the requirements for a robotic exoskeleton to assist patients with motor dysfunctions.

The first section insists on a theoretical explanation of how walking on an inclined slope differs from level ground walking. It also speaks about how inclined slope walking differs biomechanically between healthy subjects and subjects with motor dysfunctions. The next section gives an account on different tools used in motion capturing, modeling and processing. The third section focuses on currently available exoskeleton models. The next section includes the limitations of existing models and challenges expected in designing a new model.

A.2 Inclined Slope Walking

A.2.1 Phases

Unlike level ground walking, the phases of walking are different for walking up an inclined plane and walking down a declined plane.

In inclined plane walking, the body first gets ready to be pulled up during the weight acceptance phase. Next is the pull up phase where there is a progression of ascending from one step to another. The step is completed in the forward continuance phase. Next is the foot clearance phase where the leg is brought up for the next step. The final phase is the foot placement phase where the leg is placed and the second step is completed[6].

Whereas declined plane walking involves weight acceptance phase (body prepares to pull down), forward continuance phase (body starts to move forward), controlled lowering (descending from one step to another), leg pull through (leg swing through) and foot placement[6].

A.2.2 Biomechanical Differences

Walking on an inclined slope is a complex activity that requires muscle strength, coordination and body balance[2]. Physiologically, climbing up and down an inclined slope is more challenging and energy demanding than ground

(32)

level walking. This is because the body has to lift its weight against gravity[2]. The connective tissues and tendons store these gravitational potential and kinetic energies as elastic energy. This elastic energy is released as positive work done when required later[9]. The main biomechanical difference is that elastic energy stored in tendons is almost fully recovered in ground level walking. But in walking up and down an inclined slope, the level of elastic energy that can be stored is diminished. To compensate this, the body generates a net positive mechanical energy to raise the centre of mass. This is the main reason for the increase in metabolic energy cost when climbing up and down an inclined slope when compared to ground level walking. Moreover, it is estimated that this excess metabolic energy cost is compensated by the increase in net work done at the hip only, while, the performance of knee and ankle remained the same in all inclines. But it does not mean that only hip extensor muscles produce this excess power. There is also a possibility of knee extensors producing this work by contracting.

For healthy individuals, in ground level walking, the metabolic cost remains minimum when walked at an optimum speed. Whereas, it costs more when walked at a lower or higher speed. Muscle activation duration and magnitude are expected to increase in inclined plane walking which in turn causes increased metabolic cost. Due to these differences between ground level walking and walking up and down an inclined slope, the biomechanics of these movements are different at joint and muscle levels.

Subjects with motor dysfunctions such as partial spinal cord injury subjects have weak knee extensors and hip muscles that limit their walking abilities[7]. It is also dependant on other factors such as spasticity, posture and loss of proprioception. Generally, the walking pattern of SCI subjects differ from that of normal, healthy subjects in a number of ways:

• Decrease in step length

• Increase in double limb support time

• Increased knee flexion

• Abnormal knee-hip coordination

During voluntary contraction, these subjects exhibit less peak isometric torque in the muscle groups of knee and plantar flexors. Considering ankle of these subjects, there is a deficiency in propulsion that leads to inadequate push-off during walking[7].

Emilie et al. compared the gait cycle parameters of healthy and SCI subjects and found that SCI subjects walk with much less gait speed, cadence and stride length compared to healthy subjects at normal gait speed as well as slow gait speed. They also found that power peak values at ankle, hip and

(33)

knee of healthy subjects at slow gait speed were almost similar to those of SCI subjects without any much significant change. Whereas, a significant change was seen in power peak values of healthy subjects at normal gait speed compared to those of SCI subjects. These results are mainly due to motor deficit at the ankle which agrees with weaker push-off as previously discussed.

During uphill walking, healthy subjects tend to increase their step length and reduce their cadence[7]. At heel strike, there is an increase in knee flexion. During stance phase, there is an increase in leg extension. There is also an increase in propulsion, moments of hip extensor and plantar flexors. Whereas, SCI subjects have an increased hip and trunk flexion. Also, there is a weak push off at the end of stance phase[7].

During downhill walking, healthy subjects tend to decrease their step length and increase their cadence[7]. During stance phase, there is an increase in knee flexion, increase in the activity of knee extensors and dorsiflexors and breaking force. Moreover, healthy subjects tend to show a back tilt of the trunk, linearly depending on the descending slope, which provides an additional balance in downhill walking. Whereas, SCI subjects, due to constant trunk flexion position, tend to show a reduced adaptation to downhill walking.

A.3 Tools

A.3.1 Marker placement

Capturing motion data requires a motion lab equipped with camera and force plates system. The subject has to be prepared with markers placed before recording the motion. As per literature survey, 8 camera Vicon Motion Capturing System is coupled with a software Vicon Nexus which uses Plug-in Gait model that requires 23 markers placed on trunk and lower limbs of the subject[10]. Table 3.1 represents the 23 markers used and figure 3.1 shows the location of those markers.

A.3.2 Software Tools

VICON Nexus: Plug-in Gait Dynamic Model was used in Vicon Nexus motion acquisition. The model requires information of 23 markers on trunk and lower limbs[10]. Preprocessing involves automatic gap filling which corrects missing markers and ensures accuracy of collected data.

Matlab: Vicon Nexus exports motion data files as ’.c3d’ files. A special Matlab extraction tool called ’c3d2OpenSim’ tool is modified as per our requirements to extract ’.trc’ file that contains static and dynamic marker data and ’.mot’ file that contains ground reaction force data[13]. These ’.trc’ and ’.mot’ files are the inputs for the simulation in OpenSim.

(34)

Table A.1: Markers Description

Marker Description C7 7th cervical vertebra T10 10th thoracic vertebra CLAV Clavicle

STRN Sternum RBAK Right Back

LSHO/RSHO Left/Right Shoulder LASI/RASI Left/Right ASIS LPSI/RPSI Left/Right PSIS LTHI/RTHI Left/Right Thigh LKNE/RKNE Left/Right Knee LTIB/RTIB Left/Right Tibia LANK/RANK Left/Right Ankle LHEE/RHEE Left/Right Heel LTOE/RTOE Left/Right Toe

Figure A.1: Markers location (Markers used in the study are framed in red)

OpenSim: Processing in OpenSim includes scaling, inverse kinematics, inverse dynamics, residual reduction and static optimization. Scaling scales the generic musculoskeletal model to the subject dimensions. Inverse kinematics computes joint angles for the model that best produce the motion of the subject. Inverse dynamics computes net joint reaction forces and net joint moments using previously calculated joint angles, angular velocities and angular accelerations of the model and also experimental ground reaction forces and moments. Residual reduction algorithm alters the torso centre mass and allows the kinematics to be more dynamically consistent with the ground reaction force data. Static optimization works on inverse kinematics results and calculates individual muscle activation and force at each instance of time[12].

Microsoft Excel: Static optimization gives individual muscle activation in ’.sto’ file which shall be opened in MS Excel. Comparison of different muscle activation for different kinds of motions are time normalised and plotted against percent gait cycle. Plotting shall be done in OpenSim directly but MS Excel proves to be more convenient in formatting the plots.

A.4 Current trends in robotic exoskeletons

At present, robotic exoskeletons are designed and manufactured for three main areas of application: gait rehabilitation, human locomotion assistance and human strength augmentation.

A.4.1 Gait rehabilitation

Rehabilitation trainings are given to patients who lost their normal gait pattern due to neurological injuries such as stroke and spinal cord injuries.

(35)

The problem with the already existing trainings is that they required enormous man power from the therapists and also proved to be time consuming and inefficient[8]. Robotic exoskeletons replaced these conventional trainings. They provided intensive motion for the patients and their recovery could be estimated by analyzing the measured torque values by the sensors thereby enabling the therapists to concentrate more on the gait pattern analysis.

Robotic Orthosis Lokomat is a system developed by Hocoma, Switzerland. The system is composed of a robotic orthosis with four degrees of freedom. Assistive torque in knee and hip are provided by actuators. Force sensors measure these torque values. It is proved to be effective in gait rehabilitation[8].

Active Leg Exoskeleton (ALEX) is developed by Banala et al., University of Delaware. This system is improvised with seven degrees of freedom. Its effectiveness has been proved in stroke recovered patients that their gait patterns looked closer to those of healthy individuals[8].

Ekso GT is an exoskeleton developed by Ekso Bionics, USA. This system is based on ‘variable assist’ that is, it provides assistance based on the needs of the particular patient[8].

A.4.2 Human locomotion assistance

Locomotion assistance is provided for the patients who have completely lost their lower limb mobility as a result of paralysis. These exoskeletons are designed to provide external torque at joints of these patients.

The ReWalk exoskeleton developed by ReWalk Robotics, USA provides hip and knee motion power to enable SCI subjects to stand upright and walk. The device senses the upper body forward tilt and then mimics the gait pattern of a healthy individual[8].

The Vanderbilt exoskeleton built by Goldfarb et al. is the first of its kind to assist SCI patients in standing up, sitting down and walking up and down stairs. There are proven results where the hip and knee joint amplitudes are found closer to that of non-SCI subjects[8].

A.4.3 Human strength augmentation

There is a need to enhance natural ability for people like soldiers, firefighters, relief workers to perform heavy works such as carrying loads which they cannot perform with their natural ability.

(36)

Berkeley Lower Extremity Exoskeleton (BLEEX) was developed by University of California to help soldiers carry heavy loads. Results show that soldiers were able to walk at 1.3 m/s carrying a load of 34 kg.

University of Tsukuba, Japan developed Hybrid Assistive Limb (HAL) which was released in several versions such as full body, two leg and single leg versions. The full body version helped in carrying a maximum load of 70 kg.

Different exoskeletons currently available are summarized in table 1 shown below.

Figure A.2: Table summarizing currently available lower extremity exoskeletons

A.5 Limitations and challenges

The general working of an intention based robotic exoskeleton involves user’s motion data acquisition and analysis and assisting based on the user’s intention. There are different types of biomechanical data associated with human motion. They are

• kinematic data (body posture and joint angles)

• kinetic data (human joint torque, ground reaction forces, and interaction force between user and exoskeleton)

• bioelectric data (EMG signals and brain signals)

These data are acquired by different motion sensors. The user’s intention is predicted by detecting the changes occurred in these data.

(37)

Limitations include poor portability due to heavy weight. Some user movements are still unachievable. Affordability is another concern. Cost needs to be effectively reduced. Some mechanical designs alter the normal gait pattern and thereby causing high metabolic cost and discomfort to the users. The design should be customizable depending on the user’s requirements. Noise which the device makes adds to discomfort of the user and the surrounding. Though present exoskeleton technologies could achieve motions like sitting, standing and walking on a level plane, there is still a vast area to be discovered, such as, walking on an uneven surface, stepping into and out of a car and so on[8].

Future work in exoskeletons mainly aims in making the device less weight and easy to port. This requires replacing the used material with other which has low density and toughness. High efficient actuators with high power to weight ratio shall be used. Integrating motors, clutches and brakes into a single device shall be investigated. Much work is still needed in the noiseless acquisition of EMG signals from the user. Researches should be carried out in reducing the cost of the device and making the device available for everyone[8].

A.6 Summary

The literature study, so far, focused on the biomechanical differences of ground level walking and inclined slope walking, gait cycle parameters of healthy and motor impaired subjects, an overview of software tools which might be of use during the thesis project, currently available robotic exoskeletons and their pros and cons. With some more investigation on muscle activities and metabolic cost of inclined slope walking, it would be possible to design an intention based robotic exoskeleton that overcomes the existing cons and challenges.

A.7 References

[1] Patricia Krawetz, MD, Patricia Nance, MD. Gait Analysis of Spinal Cord Injured Subjects: Effects of Injury Level and Spasticity. Arch Phys Med Rehabil Vol 77, July 1996.

[2] Amanda Ortiz. The metabolic cost of walking and running up a 30 degree incline: implications for vertical kilometer foot races. Undergraduate Honors Theses. 2017

[3] Maxim N. Nikolenko, Denis A. Kotin. General Principles of Medical Exoskeleton Design. 18th International Conference on Micro/Nanotechnologies and Electrical Devices. 2017.

[4] Ian Benson, Kirsten Hart1, Dot Tussler and Joost J van Middendorp. Lower-limb exoskeletons for individuals with chronic spinal cord injury: findings from a feasibility study. Clinical Rehabilitation 2016, Vol. 30(1)

(38)

73–84.

[5] Chunjie Chen, Xinyu Wu, Du-xin Liu, Wei Feng and Can Wang. Design and Voluntary Motion Intention Estimation of a Novel Wearable Full-Body Flexible Exoskeleton Robot. Hindawi Mobile Information Systems Volume 2017.

[6] Sadiq J. Abbas and Zahraa M. Abdulhassan. Kinematic Analysis of Human Climbing up and Down Stairs at Different Inclinations. Eng. Tech.Journal, Vol. 31,Part (A), No.8, 2013.

[7] Emilie Desrosiers, Cyril Duclos and Sylvie Nadeau. Gait adaptation during walking on an inclined pathway following spinal cord injury. Clinical Biomechanics 29 (2014) 500–505.

[8] Bing Chen, Hao Ma, Lai-Yin Qin, Fei Gao, Kai-Ming Chan,Sheung-Wai Law, Ling Qin, Wei-Hsin Liao. Recent developments and challenges of lower extremity exoskeletons. Journal of Orthopaedic Translation (2016) 5, 26-37.

[9] Kotaro Sasaki, Richard R Neptune. Muscle mechanical work and elastic energy utilization during walking and running near the preferred gait transition speed. Gait Posture 23 (2006) 383–390.

[10] Vicon Motion System. Modeling with Plug-in Gait. 2018.

URL:https://docs.vicon.com/display/Nexus25/Modeling+with+Plug- in+Gaithttps://docs.vicon.com/display/Nexus25/Modeling+with+Plug-in+Gait.

[11] National Center for Simulation in Rehabilitation Research. Gait 2392 and 2354 Models. URL: https://simtk-confluence.stanford.edu/display/OpenSim/Gait+2392+and+2354+Models.

[12] National Center for Simulation in Rehabilitation Research. OpenSim User’s Guide. 2017. URL: https://simtk-confluence.stanford.edu/display/OpenSim/User%27s+Guide

[13] Felipe Costa Alvim. c3d2OpenSim. 2014. URL: https://simtk.org/projects/c3d2opensim.

(39)

B

Muscle Activity Plots

Figure B.1: Soleus muscle activation comparison (Trial 1)

Figure B.2: Soleus muscle activation comparison (Trial 2)

(40)

Figure B.3: Tibialis Posterior muscle activation comparison (Trial 1)

Figure B.4: Tibialis Posterior muscle activation comparison (Trial 1)

(41)

Figure B.5: Rectus Femoris muscle activation comparison (Trial 1)

Figure B.6: Rectus Femoris muscle activation comparison (Trial 2)

(42)

Figure B.7: Adductor Magnus muscle activation comparison (Trial 1)

Figure B.8: Adductor Magnus muscle activation comparison (Trial 1)

(43)

(44)

TRITA CBH-GRU-2019:002

Simulation of Lower Limb Muscle Activity During Inclined Slope Walking

Simulation of lower limb muscle

activity during inclined slope

walking

GANESH PRASANTH ARUMUGANAINAR

Simulation of lower limb muscle

activity during inclined slope

walking

GANESH PRASANTH ARUMUGANAINAR

Abstract

Sammanfattning

Acknowledgements

Contents

List of Figures

List of Tables

List of Abbreviations

1

Introduction

1.1

Aim

1.2

Hypothesis

2

Materials and Methods

2.1

Subject

2.2

Experimental Setup

2.3

Testing Protocol

2.4

Data Processing

3

Results

3.1

Considerations at Ankle

3.2

Considerations at Knee

3.3

Considerations at Hip

3.4

Validation of Results

4

Discussion

5

Limitations and Future Work

6

Conclusion

Bibliography

Appendices

A

Literature Study

A.1

Introduction

A.2

Inclined Slope Walking

A.3

Tools

A.4

Current trends in robotic exoskeletons

A.5

Limitations and challenges

A.6

Summary

A.7

References

B

Muscle Activity Plots