Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 42 (2014 ) 78 84 2013 International Symposium on Medical and Rehabilitation Robotics and Instrumentation Validation of Inbuilt Muscle Activation Feedback Sensor for a Closed Loop Functional Electrical Stimulator Development Norhasliza Mohamad Yusoff a, Nur Azah Hamzaid a * a Department of Biomedical Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia Abstract In this paper, we presented the first stage of our closed loop functional electrical stimulator (FES) development work in reducing errors occurred at sensory feedback during FESevoked contraction. This preliminary experimental study aims to verify that the measured voltage signal evoked by functional electrical stimulator across healthy human muscles satisfies the is independent of electromyography signal, electrostatic charges and earthing effect which has potential to affect the sensory feedback of the stimulator. Four ablebodied subjects were recruited to undergo evoked muscle contraction with voltage measured across their electrically stimulated tibialis anterior and brachialis muscles. The tolerated stimulation current amplitude used in this experiment was 20mA to 43mA. The stimulation current pulse width and frequency were fixed at 200us and 35Hz respectively, as the parameters evoked the most consistent response. Results showed that the range of measured voltage that is least affected by noise is above 0.1V at room temperature. The measured voltage satisfied current relationship (r = 0.98). A trend of increased resistance over time may was observed, which may suggest change in muscle behavior throughout the evoked contraction period. These preliminary results suggested strong relationship between muscle activation with measured voltage and resistance. This warrants further exploration study on spinal cord injured subjects with respect to force production to verify the efficacy of the sensory feedback. The acquisition region range setting will then be used on spinal cord injured subject to validate effectiveness of muscle activation feedback sensor during functional electrical stimulation assisted exercise. 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BYNCND license 2013 The Authors. Published by Elsevier Ltd. Selection and/or peerreview under responsibility of the Centre of (http://creativecommons.org/licenses/byncnd/3.0/). Peerreview Humanoid Robots under responsibility and BioSensor of (HuRoBs), the Center Faculty for Humanoid of Mechanical Robots and Engineering, BioSensing Universiti (HuRoBs) Teknologi MARA. Keywords: FES, feedback sensor, closed loop 1. Introduction Functional Electrical Stimulation (FES) is one of the clinical neurostimulation applications aimed to mimic normal voluntary movement by generating function using electrical stimulation in order to restore lost or damaged function [1]. The use of FES has been proven to work well with exercises such as cycling [24] and rowing [58]. FES with certain predetermined parameter such as current and frequency; evoked the muscles and induced contraction which results in force production at the lower and upper extremities. Coupling the FES with rowing and cycling exercises improves the control of muscle contraction especially for individuals with spinal cord injury. Despite the benefits of FESassisted exercise [9], the number of SCI individuals who benefited from this technique is still low as there has yet been a reliable closedloop gait control strategy that is widely accepted [10, 11]. Some of the closed loop control techniques * Corresponding author. Tel.: +60 3 7967 4487; fax:+60 3 7967 4579. Email address: azah.hamzaid@um.edu.my 18770509 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BYNCND license (http://creativecommons.org/licenses/byncnd/3.0/). Peerreview under responsibility of the Center for Humanoid Robots and BioSensing (HuRoBs) doi:10.1016/j.procs.2014.11.036
Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 79 proposed the use of various means of quantitative acquisition tool such as EMG [1214] and motion sensors [10]. There were also studies that proposed the use of artificial intelligent technique such as selfadaptive fuzzy logic [15, 16] and genetic algorithm [17] based on force produced during the exercise. A mathematical neuromuscularskeletal model based on sliding mode theory is another proposed closed loop control method [11]. Further study on this mathematical model was done with the combination of artificial intelligence technique or adaptive fuzzy logic [18]. Fig. 1. The research area on close loop control of FES exercise Based on Fig 1, the research area of closed loop control in FES application centered heavily around monitoring the movement output and muscle contraction using techniques that are separated from the FES device itself in order to provide feedback control to the system. As FES uses electrical signal to evoke the muscle, while treating the muscle as a load, it was hypothesized that the FES signal itself may be able to show properties of muscle contraction. However, it still remains unclear if the signal itself can be utilized as a closed loop feedback. Therefore, the aim of this paper is to establish the relationship between muscle contraction level during FES and the FES signal itself, which is represented as resistance. This requires contraction and elimination of potential noises, which may originate from electromyography signal produced by voluntary contraction, electrostatic charges and earthing effect. 2. Method Two experiments were performed in this preliminary study to identify the noises and to verify relationship. In the first experiment, voltage was measured on brachialis and tibialis anterior muscles without FES contraction. Two able bodied subjects were asked to stand while wearing shoes and barefooted when the data was collected. The second experiment was done on four able bodied subjects with FES contraction on tibialis anterior that caused the foot to dorsiflex while subjects were sitting on a chair. Stimulation parameters for this experiment were set at pulse width of 200us, frequency of 35Hz. These values were predetermined based on a preliminary experiment to identify the range of values in which the parameters result in consistent and clear response. The stimulation current amplitude was determined by a value that can be tolerated by the subject at each time data was taken, hence it was increased over the time of the experiment as the subject gets used to the FES contraction. NI6008 data acquisition card was used to measure the voltage using 32mm diameter electrodes for both experiments.
80 Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 Ottobock Stiwell stimulator was used for the second experiment with the same electrode size. The electrodes placement was illustrated in Fig 2. Fig 2. Stimulating (red) and voltage measuring (blue) electrode positioning The signal pattern was observed and the data was collected for further analysis using MATLAB. In the first experiment, the maximum voltage and highest voltage occurrence in 30 seconds time frame were recorded while in the second experiment, the data was collected for one minute. The signals were isolated from the noises detected from the first experiment and the highest voltage occurrence during s resistance was calculated and the correlation coefficient was determined. 3. Results and Discussion 3.1. Experiment 1: Selecting measured voltage acquisition range The signal pattern was observed and it showed obvious differences between signals measured while barefooted and with shoes, i.e. noninsulated vs insulated condition. Fig 3 shows the signal pattern that was then presented in histogram to indicate the voltage distribution. The pattern was similar across all subjects. No of Occurrence (a) Time (s) (b)
Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 81 No of Occurrence (c) Time (s) (d) Fig 3. (a) Measured signal while wearing shoes (b) Histogram of measured signal while wearing shoes (c) Measured signal while barefooted (d) Histogram of measured signal while barefooted Signal pattern revealed higher measured voltage which occurred throughout a greater range when the subject was barefooted. This agrees with the potential effect of earthing and electrostatic charges, thus a voltage range needed to be specified to isolate the noises. Further analysis was performed to reveal that the measured voltage without stimulated contraction occurred around 0.1V and below (Table 1). Table 1. Highest voltage occurrence and maximum voltage measured in a 30 seconds time frame without FES contraction. Experiment 1 Subject 1 Subject 2 Subject 3 Subject 4 Location A A B A Highest voltage occurrence (V) Wearing shoes Barefooted 0.006 0.009 Maximum Voltage Measured (V) Wearing shoes 0.019 0.019 0.021 0.019 0.029 0.011 0.024 0.024 Barefooted 0.085 0.098 0.059 0.098 0.067 0.103 0.042 0.047 * note that the values are all below 0.1 V Given the hardware limitation of data acquisition card that can only detect 10V and below, the range of measured voltage during FES contraction acquisition region is illustrated in Fig 4.
82 Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 3.2. Experiment 2: Verifying ohm s law Fig 4: Measured voltage during FES contraction acquisition region Resistance value was calculated based on the measured voltage and the stimulation current (Fig 5). Results indicated highly linear relationship between the measured voltage and the stimulation current (r = 0.98) and a weak correlation between resistance and stimulation current (r = 0.37). Fig 5(a) illustrates the increased resistance with increased current. The increase in current throughout the session was due to the better toleration among subjects in a particular session. However, the resistance value is expected to be constan increment might indicate the early signs of physiological changes that might occur over a stimulation period. These changes have been suggested, to some extent, to cause fatigue that shortens the exercising time for spinal cord injured (SCI) individuals during training. These findings warrants an exploration study on SCI subjects and validation on efficacy of using measured voltage of FES signal as inbuilt muscle activation feedback sensor with the SCI as target population. (a)
Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 83 (b) 4. Conclusion Fig 5. (a) Resistance vs stimulating current graph with r = 0.37 (b) Measured voltage vs stimulating current graph with r = 0.98 This study successfully verified that the measured voltage signal across a healthy human muscle evoked by functional electrostatic charges and earthing, which may otherwise affect the sensory feedback of the stimulator, could be isolated. Further investigation should be conducted to verify if the current behaves similarly across atrophied SCI muscles when being electrically evoked. Acknowledgements The authors gratefully acknowledge the financial support by the University of Malaya (UM), Malaysia to carry out this research project. The project was funded under the project UM.C/HIR/MOHE/ENG/39. References [1] Rushton, D.N., Functional electrical stimulation. Physiol Meas, 1997. 18(4): p. 24175. [2] Hasnan, N., et al., Exercise Responses during FES Cycling in Individuals with Spinal Cord Injury. Med Sci Sports Exerc, 2012. [3] Castello, F., et al., The use of functional electrical stimulation cycles in children and adolescents with spinal cord dysfunction: a pilot study. J Pediatr Rehabil Med, 2012. 5(4): p. 26173. [4] Brurok, B., et al., Effect of lower extremity functional electrical stimulation pulsed isometric contractions on arm cycling peak oxygen uptake in spinal cord injured individuals. J Rehabil Med, 2013. 45(3): p. 2549. [5] Wheeler, G.D., et al., Functional electric stimulationassisted rowing: Increasing cardiovascular fitness through functional electric stimulation rowing training in persons with spinal cord injury. Arch Phys Med Rehabil, 2002. 83(8): p. 10939. [6] Hettinga, D.M. and B.J. Andrews, Oxygen consumption during functional electrical stimulationassisted exercise in persons with spinal cord injury: implications for fitness and health. Sports Med, 2008. 38(10): p. 82538. [7] Jeon, J.Y., et al., Reduced plasma glucose and leptin after 12 weeks of functional electrical stimulationrowing exercise training in spinal cord injury patients. Arch Phys Med Rehabil, 2010. 91(12): p. 19579. [8] Taylor, J.A., G. Picard, and J.J. Widrick, Aerobic capacity with hybrid FES rowing in spinal cord injury: comparison with armsonly exercise and preliminary findings with regular training. PM R, 2011. 3(9): p. 81724. [9] Hamzaid, N.A. and G.M. Davis, Health and fitness benefits of functional electrical stimulationevoked leg exercise for spinal cordinjured individuals:a position review. Top. Spinal Cord Inj.Rehabil, 2009. 14: p. 33. [10] Braz, G.P., et al., Motion Sensors Feedback in FES Gait: A Novel Control Strategy, in World Congress on Medical Physics and Biomedical Engineering 2006, R. Magjarevic and J.H. Nagel, Editors. 2007, Springer Berlin Heidelberg. p. 28602863.
84 Norhasliza Mohamad Yusoff and Nur Azah Hamzaid / Procedia Computer Science 42 ( 2014 ) 78 84 [11] Jezernik, S., R.G. Wassink, and T. Keller, Sliding mode closedloop control of FES: controlling the shank movement. IEEE Trans Biomed Eng, 2004. 51(2): p. 26372. [12] Mizrahi, J., et al., EMG as an indicator of fatigue in isometrically FESactivated paralyzed muscles. Rehabilitation Engineering, IEEE Transactions on, 1994. 2(2): p. 5765. [13] Chesler, N.C. and W.K. Durfee, Surface EMG as a fatigue indicator during FESinduced isometric muscle contractions. J Electromyogr Kinesiol, 1997. 7(1): p. 2737. [14] Solomonow, M., et al., The Myoelectric Signal of Electrically Stimulated Muscle During Recruitment: An Inherent Feedback Pareter for a Closed Loop Control Scheme. Biomedical Engineering, IEEE Transactions on, 1986. BME33(8): p. 735745. [15] Davoodi, R. and B.J. Andrews, Computer simulation of FES standing up in paraplegia: a selfadaptive fuzzy controller with reinforcement learning. IEEE Trans Rehabil Eng, 1998. 6(2): p. 15161. [16] Davoodi, R. and B.J. Andrews, Fuzzy logic control of FES rowing exercise in paraplegia. Biomedical Engineering, IEEE Transactions on, 2004. 51(3): p. 541543. [17] Davoodi, R. and B.J. Andrews, Optimal control of FESassisted standing up in paraplegia using genetic algorithms. Medical Engineering & Physics, 1999. 21(9): p. 609617. [18] Nekoukar, V. and A. Erfanian, An adaptive fuzzy slidingmode controller design for walking control with functional electrical stimulation: A computer simulation study. International Journal of Control, Automation and Systems, 2011. 9(6): p. 11241135.