An Inertial Sensor-based Trigger Algorithm for Functional Electrical Stimulation-Assisted Swimming in Paraplegics

2nd IFAC Conference on Cyber-Physical & Human-Systems 2nd Conference on & 2nd IFAC IFAC on Cyber-Physical Cyber-Physical & Human-Systems Human-Systems Miami, FL,Conference USA, Dec. 14-15, 2018 2nd IFAC Conference on Cyber-Physical & Human-Systems Miami, FL, USA, Dec. 14-15, 2018 Available online at www.sciencedirect.com Miami, FL, USA, Dec. 14-15, 2018 2nd IFAC on Cyber-Physical & Human-Systems Miami, FL,Conference USA, Dec. 14-15, 2018 Miami, FL, USA, Dec. 14-15, 2018

ScienceDirect

IFAC PapersOnLine 51-34 (2019) 278–283

An An An An

Inertial Inertial Sensor-based Sensor-based Trigger Trigger Algorithm Algorithm Inertial Sensor-based Trigger Algorithm for Functional Inertial Trigger Algorithm forSensor-based Functional Electrical Electrical for Functional Electrical Stimulation-Assisted Swimming for Functional Electrical Stimulation-Assisted Swimming in in Stimulation-Assisted Swimming in Paraplegics Stimulation-Assisted Swimming in Paraplegics Paraplegics Paraplegics Constantin Wiesener ∗∗ Thomas Seel ∗∗ Jens Axelgaard ∗∗ ∗∗

∗ Thomas Seel ∗ Jens Axelgaard ∗∗ Constantin Wiesener Constantin Wiesener Seel∗∗∗ Axelgaard ∗∗ ∗ Thomas ∗ Jens ∗∗ ∗ Rachel Horton Niedeggen Thomas Schauer ∗∗ ∗ Constantin Wiesener Thomas Seel∗∗∗ Jens Axelgaard ∗∗ Andreas ∗∗∗ Rachel Horton Andreas Niedeggen Schauer ∗ ∗ Thomas ∗∗ ∗ Rachel Horton Andreas Niedeggen Thomas Schauer ∗∗ ∗∗∗ Constantin Wiesener Thomas Seel Jens Axelgaard Rachel Horton Andreas Niedeggen Thomas Schauer ∗∗ ∗∗ ∗∗∗ ∗ Rachel Horton Andreas Niedeggen Thomas Schauer a t Berlin, Berlin, ∗ Control Systems Group at Technische Universit¨ ∗ Control Systems Group Technische a Systems (e-mail: Group at [email protected]). Technische Universit¨ Universit¨ att Berlin, Berlin, Berlin, Berlin, ∗ ControlGermany Control Systems Group at Technische Universit¨ a t Berlin, Berlin, Germany (e-mail: [email protected]). ∗ (e-mail: [email protected]). ∗∗ ControlGermany Systems Group at Technische Universit¨ a t Berlin, Berlin, Manufacturing Co., Ltd., USA ∗∗ Germany (e-mail: [email protected]). ∗∗ Axelgaard Axelgaard Manufacturing Co., Ltd., Ltd., USA USA Axelgaard Manufacturing Co., ∗∗∗ ∗∗Centre Germany (e-mail: [email protected]). Treatment for Spinal Cord Injuries, ukb Unfallkrankenhaus ∗∗∗ Axelgaard Manufacturing Co., Ltd., USA ∗∗∗ Treatment∗∗Centre for Cord Injuries, ukb for Spinal Spinal Cord Injuries, ukb Unfallkrankenhaus Unfallkrankenhaus ∗∗∗ Treatment Centre Axelgaard Manufacturing Co., Ltd., USA Berlin, Germany Treatment Centre for Spinal Cord Injuries, ukb Unfallkrankenhaus Berlin, Germany ∗∗∗ Berlin,Cord Germany Treatment Centre for Spinal Injuries, ukb Unfallkrankenhaus Berlin, Germany Berlin, Germany Abstract: Functional electrical stimulation (FES) is used to support gait in stroke patients and Abstract: Functional electrical stimulation (FES) is is used used to to support support gait gait in in stroke stroke patients patients and and Abstract: Functional electrical stimulation (FES) to induce cycling motions in paralyzed legs. In the current contribution, we present a method Abstract: Functional electrical stimulation (FES) is used to support gait in stroke patients and to induce cycling motions in paralyzed legs. In the current contribution, we present a method to induce cycling motions in paralyzed legs. In the current contribution, we present a method Abstract: Functional electrical stimulation iscurrent used toin support gait in stroke patients and that, for the first time, enables FES-supported swimming paraplegics. The proposed setup to induce cycling motions in paralyzed legs.(FES) In the contribution, we present a method that, for first time, enables FES-supported swimming in paraplegics. The proposed setup that, for athe the first motions time,stimulator, enables FES-supported swimming in paraplegics.we The proposed setup to induce cycling in paralyzed legs. In electrodes. the currentIn contribution, present a method includes waterproof cables, and preliminary experiments, flexion that, for the first time, enables FES-supported swimming in paraplegics. The proposed setup includes waterproof stimulator, cables, and electrodes. In preliminary experiments, flexion includes waterproof cables, andgenerated electrodes. In preliminary experiments, flexion that, for aa first time,stimulator, enables FES-supported swimming in paraplegics.paralyzed The proposed setup and extension movements of the knee were in aa completely subject to includes athe waterproof stimulator, cables, andgenerated electrodes. In preliminary experiments, flexion and extension movements of the knee were in completely paralyzed subject to and extension movements of the knee were generated in a completely paralyzed subject to includes a waterproof stimulator, cables, and electrodes. In preliminary experiments, flexion support the propulsion. Furthermore, transcutaneous spinal cord stimulation (tSCS) is used and extension movements of the knee were generated in a completely paralyzed subject to support the propulsion. Furthermore, transcutaneous spinal cord stimulation (tSCS) is used support the propulsion. Furthermore, transcutaneous spinal cord stimulation (tSCS) is used and extension movements of the knee were generated in a completely paralyzed subject to to get a straight swimming position and to reduce spasticity of the lower extremities. The support the propulsion. Furthermore, transcutaneous spinal cord stimulation (tSCS) is used to get a straight swimming position and to reduce spasticity of the lower extremities. The to get a the straight swimming position and to reduce spasticity of the lower extremities. The support propulsion. Furthermore, transcutaneous spinal cord stimulation (tSCS) is used developed setting remained dry and safe during all sessions. The first trials revealed the need for to get a straight swimming position and to reduce spasticity of the lower extremities. The developed setting dry and safe during all sessions. The trials revealed the developed setting remained remained dry and safe during allwith sessions. The first first trials revealed the need need for for to get a straight swimming position and to reduce spasticity of the lower extremities. a synchronization of the patient’s arm movements the artificially induced leg movements to setting remained dry and safe during allwith sessions. The first trials revealed the needThe for adeveloped synchronization of the patient’s arm movements the artificially induced leg movements to a synchronization of the patient’s arm movements with the artificially induced leg movements to developed setting remained dry and safe during all sessions. The first trials revealed the need for prevent undesired rolling movements of the swimmer. To enable such a synchronized swimming, a synchronization of the patient’s arm movements with the artificially induced leg movements to prevent undesired rolling movements of the swimmer. To enable such a synchronized swimming, prevent undesired rolling movements of the swimmer. To enable such a synchronized swimming, a synchronization of the patient’s arm with the artificially legvelocity movements to a trigger algorithm was developed that is based on the roll angle angular of the undesired rolling movements ofmovements the swimmer. To enable suchand ainduced synchronized swimming, aprevent trigger algorithm was developed that is based on the roll angle and angular velocity of the a trigger algorithm was validation, developed is demonstrated based on the roll angle and angular velocity of the prevent undesired rolling movementsthat ofwas the swimmer. To enable such a synchronized swimming, trunk. By experimental it that a functional stimulation pattern a trigger algorithm was developed that is based on the roll angle and angular velocity of the trunk. By experimental it was that aa functional stimulation pattern trunk. By experimental validation, it was demonstrated thatbody. functional stimulation pattern a trigger algorithm was validation, developed that is demonstrated basedofon the roll angle and new angular velocity of the can be generated during front crawl movements the upper The setup and methods trunk. By experimental validation, it was demonstrated thatbody. a functional stimulation pattern can be generated during front crawl movements of the upper The new setup and methods can be generated during front crawl movements of the upper body. The new setup and methods trunk. By experimental validation, it was demonstrated that a functional stimulation pattern are being tested during the STIMSWIM study with paraplegics. The preliminary can currently be generated during front crawl movements of pilot the upper body. The new setup and methods are currently being tested during the STIMSWIM study with The preliminary are currently being tested during themovements STIMSWIM pilot study with paraplegics. paraplegics. The preliminary can be generated during front crawl of pilot the upper body. The new setup and methods results of the first two subjects show an improvement of the swimming speed of approximately are currently being tested during the STIMSWIM pilot study with paraplegics. The preliminary results of the first two subjects show an improvement of the swimming speed of approximately results ofFES/tSCS the being first two subjects show ancompared improvement ofstudy the swimming speed ofThe approximately are currently tested during the STIMSWIM pilot with paraplegics. preliminary 15% for assisted swimming to non-assisted swimming and a clear training results of the first two subjects show an improvement of the swimming speed of approximately 15% assisted swimming to swimming and clear 15% for for FES/tSCS assisted swimming compared to non-assisted non-assisted swimming andofaaapproximately clear training training results ofFES/tSCS the first two subjects show ancompared improvement of the swimming speed effect over the first 7 sessions. 15% for FES/tSCS assisted swimming compared to non-assisted swimming and a clear training effect over the 77 sessions. effect over the first first assisted sessions. 15% for FES/tSCS swimming compared to non-assisted swimming and a clear training effect over the first 7 sessions. © 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. effect over the first 7 sessions. Keywords: Neurostimulation; Assistive devices Keywords: Keywords: Neurostimulation; Neurostimulation; Assistive Assistive devices devices Keywords: Neurostimulation; Assistive devices Keywords: Neurostimulation; Assistive devices 1. INTRODUCTION In the majority of cases, SCI results in complete or incom1. INTRODUCTION INTRODUCTION In the the majority majority of of cases, cases, SCI SCI results results in in complete complete or or incomincom1. In plete paralysis of the lower extremities. Therefore, effective 1. INTRODUCTION In the majority of cases, SCI results in complete or incomplete paralysis of the lower extremities. Therefore, effective plete paralysis of the lower extremities. Therefore, effective 1. INTRODUCTION In the majority ofthe cases, SCI resultsisinlimited complete ortraining incomand safe lower extremity training and plete paralysis lower extremities. Therefore, and safe lower of extremity training is limited limited and effective training and lower training is and training pletesafe paralysis ofextremity the lower extremities. Therefore, effective exercises of the upper extremities are recommended such and safe lower extremity training is limited and training exercises of the the extremity upper extremities extremities are recommended such exercises of upper are recommended such andthe safe lower training is limited and or training A spinal cord injury (SCI) is often associated with paralas use of arm-crank or wheelchair ergometer swimexercises of the upper extremities are recommended such A spinal spinal cord cord injury injury (SCI) (SCI) is is often often associated associated with with paralparal- as as the use of arm-crank or wheelchair ergometer or swimA the use of arm-crank or wheelchair ergometer or swimexercises upper extremities are recommended such ysis of the lower extremities which means a severe severe restricA spinal injury (SCI) iswhich often means associated withrestricparal- as ming. exercises can improve physical fitness by the All useofthese ofthe arm-crank wheelchair or swimysis of the thecord lower extremities ming. All these exercisesor can improve ergometer physical fitness fitness by ysis of lower extremities means aa severe ming. All these exercises can improve physical by A spinal cord injury (SCI) iswhich often associated withrestricparalas the use of arm-crank or wheelchair ergometer or swimtion of physical activity and health for the affected subysis of the lower extremities which means a severe restricup to 25 % with regular exercises (Nash, 2005). ming. All these exercises can improve physical fitness by tion of physical activity and health for the affected subup to 25 % with regular exercises (Nash, 2005). tion of physical activity and health for the affected subup to 25 % with regular exercises (Nash, 2005). ysis of the lower extremities which means a severe restricming. All these exercises can improve physical fitness by jects. on the level severity of the injury, this tion ofDepending physical activity andand health for the affected to 25 %inwith regular exercises (Nash, 2005). jects. Depending on the the level level and severity of the the injury,subthis up jects. Depending on and severity of injury, this Mobility the water is often the only experience of tion of physical activity and health for the affected subup to 25 % with regular exercises (Nash, 2005). involves a functional limitation of various body sensory Mobility in in the the water water is is often often the the only only experience experience of of jects. Depending on thelimitation level and severity of the injury, this Mobility involves a functional of various body sensory involves a functional limitation of various body sensory unaided body movement (except for the transfer in and in the water is(except often the onlytransfer experience of jects.motor Depending on the levelthe andlevel severity of the injury, this and functions below of lesion. lesion. In case of a Mobility unaided body movement for the the in and and involves a functional limitation of various body sensory unaided body movement (except for transfer in and motor functions below the level of In case of a Mobility in the water is often the only experience of and motor functions below the level of islesion. In case of a unaided from the pool) within an environment most paralyzed movement (except for that the transfer in and involves a SCI, functional limitation of various body sensory traumatic the physical inactivity in stark stark contrast from the body pool) within within an environment environment that most paralyzed paralyzed and motor functions below the level of islesion. In case of a from the pool) an that most traumatic SCI, the physical inactivity in contrast unaided body movement (except for the transfer in and traumatic SCI, the physical inactivity is in stark contrast patients enjoy. In addition, there is aa plurality of the pool) an environment most paralyzed andthe motor functions below theinjury, level ofespecially Infor case of a from to condition prior to the young patients enjoy.within In addition, addition, there is that plurality of therthertraumatic SCI, the physical islesion. in stark contrast patients enjoy. In there is a plurality of therto the condition condition prior to the theinactivity injury, especially especially for young from theeffects pool) within an environment that most paralyzed to the prior to injury, for young apeutic of swimming for paraplegics described in patients enjoy. In addition, there is a plurality of thertraumatic SCI, the physical inactivity is in stark contrast patients. apeutic effects of swimming for paraplegics described in to the condition prior to the injury, especially for young apeutic effects of swimming for paraplegics described in patients. patients enjoy. In addition, there is a plurality of therpatients. the literature as an increase in muscle strength, improved apeutic effects of swimming for paraplegics described in to the condition prior to the injury, especially for young the the literature asofan answimming increase in informuscle muscle strength, improved patients. literature as increase strength, improved apeutic effects paraplegics described in Participation in physical and therapeutic activities coordination, reduction of spasticity aa reduction of literature as an increase in muscle and strength, improved patients. Participation in physical physical and and therapeutic therapeutic activities activities followfollow- the coordination, reduction of spasticity spasticity and reduction of Participation in followcoordination, reduction of and a reduction of the literature as an increase in muscle strength, improved ing a paraplegia is often limited due to the loss of voluntary Participation in is physical and therapeutic activities follow- coordination, contractures (Bromley, 2006). reduction of spasticity and a reduction of ing a paraplegia often limited due to the loss of voluntary contractures (Bromley, 2006). ing a paraplegia is often limited due to the loss of voluntary contractures (Bromley, 2006). Participation in and physical and therapeutic followcoordination,(Bromley, reduction2006). of spasticity and a reduction of motor function inefficient temperature of ing a paraplegia often limited due to the activities lossregulation of voluntary motor function is and inefficient temperature regulation of contractures motor function and inefficient temperature of Functional electrical stimulation (FES) is used successfully ing aaffected paraplegia is often limited due to the lossregulation of voluntary contractures (Bromley, 2006). the extremities, autonomic dysfunction, and early Functional electrical stimulation (FES) is is used used successfully successfully motor function and inefficient temperature regulation of Functional electrical stimulation (FES) the affected extremities, autonomic dysfunction, and early early the affected extremities, autonomic dysfunction, and in cycling or rowing (Newham and Donaldson, 2007; Functional electrical stimulation (FES) is used successfully motor function and inefficient temperature regulation of muscle fatigue. In addition, often specially adapted equipin cycling or rowing (Newham and Donaldson, 2007; the affected extremities, autonomic dysfunction, and early in cycling or rowing (Newham and Donaldson, 2007; muscle fatigue. In addition, addition, often specially specially adapted equipFunctional electrical stimulation (FES) isDonaldson, used successfully muscle fatigue. In often adapted equipGibbons et al., 2016; Wiesener and Schauer, 2017; Schauer, in cycling or rowing (Newham and 2007; the affected extremities, autonomic dysfunction, and early ment and assistants are needed. Despite all these obstaGibbons et al., 2016; Wiesener and Schauer, 2017; Schauer, often specially adapted equipmuscle fatigue. In addition, Gibbons et al., 2016; Wiesener and Schauer, 2017; Schauer, ment and assistants are needed. Despite all these obstain cycling or 2016; rowing (Newham and Donaldson, 2007; ment and assistants are needed. Despite all these equipobsta2017). The corresponding muscles for knee extension and et al., Wiesener and Schauer, Schauer, muscle fatigue. In therapeutic addition, often specially adapted cles, sportive and activity after paraplegia can Gibbons 2017). The corresponding muscles for knee knee 2017; extension and ment and assistants are needed. Despite these obsta2017). The corresponding muscles for extension and cles, sportive and therapeutic therapeutic activity afterall paraplegia can Gibbons etwell al., 2016; Wiesener and Schauer, 2017; Schauer, cles, sportive and activity after paraplegia can flexion as as hip extension are stimulated depending 2017). The corresponding muscles for knee extension and ment and assistants are needed. Despite all these obstacontribute to a reduction in concomitant diseases and to flexion as well as hip extension are stimulated depending cles, sportive and therapeutic activity after paraplegia can flexion as well as hip extension are stimulated depending contribute to and reduction in concomitant concomitant diseases and to to on 2017). The corresponding muscles kneeorextension and contribute to aathe reduction in diseases and the or joint during cycling triggered by as well hip angle extension arefor stimulated depending cles,increase sportive therapeutic activity after paraplegia an emotional well-being of those affected on the crank crank oras joint angle during cycling or triggered triggered by contribute toin athe reduction in concomitant diseases andcan to flexion on the crank or joint angle during cycling or by an increase in emotional well-being of those affected flexion as well as hip extension are stimulated depending an increase in the emotional well-being of those affected a pull switch while rowing. Due to the combination of arm on the crank or joint angle during cycling or triggered by contribute to a reduction in concomitant diseases and to (Tawashy et al., 2009; Brunelli, 2014). a pull switch while rowing. Due to the combination of arm an increaseet in emotional well-being rowing. to the combination of arm (Tawashy al.,the 2009; Brunelli, 2014). of those affected aonpull theswitch crank while or joint angle Due during cycling or triggered by (Tawashy al., 2009; Brunelli, 2014). switch while rowing. Due to the combination of arm an increaseet emotional well-being (Tawashy et in al.,the 2009; Brunelli, 2014). of those affected aa pull pull switch while rowing. Due to the combination of arm (Tawashy et al., 2009; Brunelli, 2014). Copyright ©2018 334Hosting by Elsevier Ltd. All rights reserved. 2405-8963 © 2019, IFAC IFAC (International Federation of Automatic Control)

Copyright ©2018 IFAC 334 Copyright 334Control. Peer review©2018 under IFAC responsibility of International Federation of Automatic Copyright ©2018 IFAC 334 10.1016/j.ifacol.2019.01.039 Copyright ©2018 IFAC 334

IFAC CPHS 2018 Miami, FL, USA, Dec. 14-15, 2018

Constantin Wiesener et al. / IFAC PapersOnLine 51-34 (2019) 278–283

279

Stimulator with 4 channels in a waterproof diving bag and a customised swimming firmware

IMU sensors with data storage

Periodic stimulation of quadriceps (25 Hz, 0-500 us, 0-100 mA, ON/OFF: 0.5/0.5s) Permanent tSCS (T11/12) (50 Hz, 1 ms, 40 mA — sensory level)

Floats at the shank

Fig. 1. The world’s first FES swimming system including a waterproof stimulator, waterproof IMU sensors and waterproof electrodes. and leg training, a significantly higher training effect can be achieved. In addition, an improvement in perfusion and lower limb bone density has been observed in some studies (Newham and Donaldson, 2007). To our best knowledge, there is no prior work on FES assisted swimming for paraplegics or healthy subjects. In this paper, we present our recent work on FES assisted swimming in paraplegic patients. At first, the overall experimental setup, the swimming style, the stimulation pattern and preliminary results with one paraplegic subject are presented. Afterwards, the roll angle-based approach for triggering the electrical stimulation to produce an arm-synchronous leg movement is described. Finally, the preliminary results of the STIMSWIM study are presented and discussed.

2. THE FES SWIMMING SYSTEM The first FES swimming system was developed (Fig.1) .The stimulator (RehaMove3, Hasomed GmbH, Germany) has a customized firmware and is placed inside a waterproof bag under the swimmer’s T-shirt. For data logging, two inertial motion units (IMU) (MuscleLab, Ergotest Innovation AS, Norway) with internal data storage have been attached between the shoulder blades and at the right arm also using waterproof bags. Due to the fact that chlorinated water in swimming pools has a conductance of 2.5-3 mS/cm which results in resistance of 333-400 Ohm, a direct stimulation with non-waterproof electrodes would produce a parasitic short circuit between electrodes during stimulation. Therefore, Axelgaard Manufacturing developed a waterproof electrode with a snap connector (see Fig. 3). To fix the snap connector, a transparent film dressing (3M Tegaderm, 3M Co., USA) has been used. In tests with healthy subjects, it has been shown that the connection between cable and electrode must be waterproof as well. Otherwise, parasitic short circuits occur. Therefore, a removable tight silicone tube is used as a cover for the connection between electrode lead and cable. 335

Fig. 2. Photo and construction plan of Axelgaard R Ultrastim snap electrode with oversize water fast backing with an electrode area of 22.9 cm2 (Axelgaard, 2004, 2010) 3. SWIMMING STYLE AND SUPPORTED LEG MUSCLES The first and most natural question to ask is which leg motion can be and should be induced by FES during what style of swimming. In conventional swimming therapy

IFAC CPHS 2018 280 Miami, FL, USA, Dec. 14-15, 2018

Constantin Wiesener et al. / IFAC PapersOnLine 51-34 (2019) 278–283

for paraplegics, the independence in the water is first achieved on the back since the prone position is more difficult to maintain without muscular control of the hip joints. During the relearn phase, the swimming instructor teaches the patient how to perform precise symmetrical strokes, because asymmetrical strokes easily cause the paralyzed limbs to roll and make it difficult to maintain a straight course (Bromley, 2006). Subsequently, training may focus on several possible swimming techniques for paraplegics with only slight modifications compared to swimming styles of non-paralyzed subjects. For patients with thoracic or lumbar lesion height, Bromley (2006) recommends backstroke, breaststroke, crawl stroke and butterfly stroke, among which the butterfly stroke is the most difficult to learn. For normal breaststroke in unimpaired swimmers, the socalled frog kick is used as leg technique, which includes knee flexors and extensors, thigh adductors and abductors, gluteus and the plantar-flexors. In preliminary tests, we found out that a complex movement like the frog kick is currently not realizable with FES due to the high number of involved muscle groups. For backstroke and crawl, the so-called flutter kick can be used, which involves mostly the knee extensors and flexors as well as the gluteus muscles. In (Seifert et al., 2011, 2010) the knee angle course for healthy non-expert swimmers for crawl is analyzed. Here, after a short and strong extension phase, a plateau phase can be observed where the knee joint is fully extended. During the plateau phase, the other knee is flexed to 40-50 degrees and then immediately and quickly extended and then enters the plateau phase. To find the best swimming technique and stimulation setting, several preliminary tests have been executed with a paraplegic subject (ASIA impairment scale A, lesion level T5/6) under medical supervision of the Unfallkrankenhaus Berlin 1 . The participant gave written informed consent. The stimulation of the gluteus was excluded since a paraplegic would not be able to place the electrodes on that muscle without assistance. Furthermore, the hip position of a paraplegic in backstroke swimming technique depends on the level of control over the waist and hip. In preliminary tests, we found out that the lower the hip is, the less propulsion can be achieved by stimulating the knee flexors and extensors. Therefore, we decided to use crawl strokes in combination with FES-induced flutter kicks for the planned study. Furthermore, preliminary tests revealed that attaching floats to the ankle joints leads to a default upward movement of the ankle, which results in a more streamlined position in the water. In Fig. 3 three photos in the sagittal plane are displayed which show the different states of the stimulation. During this test, the subject was asked not to swim with his arms to get a stable position. During the trial underwater video data has been captured using a waterproof camera. During all trials, the influence of the Hamstring stimulation was rated quite low compared to the propulsion produced by the quadriceps. After an extension, the knee automatically flexes itself back if floats at the ankle are used. Hence, during the subsequently reported trials only the quadriceps muscle groups were stimulated periodically for 0.5 s 1

Ethical approval of Berlin Chamber of Physicians Eth-28/17

336

Rest / No stimulation

Extension of right leg with FES

Extension of left leg with FES Fig. 3. Experiment with a paraplegic subject (Th5, ASIA scale A) with floats at the feet with antisynchronous stimulation of the left and right quadriceps. at 25 Hz with a adjustable pulse width of 0-500 µs and current amplitude of 0-100 mA. Furthermore, transcutaneous spinal cord stimulation (tSCS) was applied at a level of 40 mA and 1 ms pulse width at a frequency of 50 Hz over the T11/12 region at the spinal cord (Hofstoetter et al., 2014). The level was chosen to produce a sensory stimulation of the lower extremity nerves for spasticity reduction. At the same time, the trunk musculature is activated at a motor level by the tSCS. The latter shall improve trunk stability and straighten the upper body and hip. Analyzing the video data and comparing the knee angles to Seifert et al. (2011, 2010) the plateau phase of the knee angle after the full extension is reduced and due to the floats and the used stimulation pattern. The knee moves automatically to the rest angle of 90 degree. Furthermore, the reached minimum flexion angles are 20 degrees higher compared to non-paralyzed swimmers. Using the IMU sensors the roll angle of the upper trunk and the right arm inclination angle were measured during the swimming experiments, as presented in Fig. 4. According to Psycharakis and Sanders (2010), the measured or observed roll angle is in the range of non-expert swimmers. Furthermore, the arm inclination angle is synchronous to the upper trunk roll angle. From video analysis, the trunk and hip roll angle performs similar, since no synchronization of arm and leg movement is performed. Psycharakis and Sanders (2010) described an decrease of the hip roll angle compared to the upper trunk roll angle with increasing swimming velocity and with synchronous arm and leg

IFAC CPHS 2018 Miami, FL, USA, Dec. 14-15, 2018

Constantin Wiesener et al. / IFAC PapersOnLine 51-34 (2019) 278–283

281

zE 60

40

Angles (degrees)

20

0

yE φ

-20

-40

yS

-60

-80 50

Roll Angle Inclination of the Right Arm

51

52

53

54

55

56

57

58

59

60

Time (s)

Fig. 4. Roll angle of the upper trunk and inclination angle of the right arm of a paraplegic subject during FES assisted front crawl swimming. The inclination angle of the arm is defined in the sagittal plane where 0 ◦ means that the arm is parallel to the water surface and points forward, negative angle means the arm is pointing downwards and a positive angle means the arm is pointing upwards. movement. Therefore, a major result of these preliminary experiments is that, in the case of crawl stroke, the knee extension should be synchronized with the contralateral arm movement to increase the swimming speed and effectiveness. Beside the experimental results for the swimming style the stimulation setting including the stimulator, cables, and electrodes stayed dry over the full trial duration of 30 minutes in all experiments. 4. ROLL-ANGLE-TRIGGERED STIMULATION To address the need for synchronization of the leg movements with the arm movements, a method was developed that uses the roll angle and angular velocity of the trunk to trigger the stimulation of the legs. The roll angle of the trunk is defined as the angle between the mediolateral axis of the trunk and the horizontal plane, as illustrated in Fig. 5. During front crawl swimming, this angle typically varies periodically in the range of ±50 ◦ for expert swimmers and in the range of ±30 ◦ for non-experts swimmers (cf. (Callaway et al., 2009; B¨ achlin et al., 2009)). An IMU is attached to the back such that the intrinsic x-axis of the IMU is aligned with the longitudinal axis of the trunk and the intrinsic y-axis of the IMU points mediolaterally to the right-hand side of the subject. Therefore, the roll angle can be calculated via the following procedure. In a first step, the orientation of the IMU is determined from the measured acceleration and angular rate. To this end, we employ the algorithm proposed by Seel and Ruppin (2017), which yields a quaternion SE q[k] at every time index k. This quaternion describes the orientation of the intrinsic sensor frame S of the IMU with respect to a fixed inertial reference frame E with vertical z-axis. Note that the algorithm is modular and that we refrain from using the magnetometer-based corrections described by Seel and Ruppin (2017) information since the azimuth (heading) information is not needed. The y-axis of the IMU in sensor coordinates is given by SS y = [0 1 0]T . The reference frame coordinates of this vector are obtained by 337

Fig. 5. Roll angle φ of the trunk where zE is the z-axis of the world frame and zS and yS are the z- and y-axis of the intrinsic sensor frame.     0 0 S = E q[k] ⊗ S ⊗ SE q −1 [k], (1) S y E Sy where ⊗ denotes quaternion multiplication. Likewise, the measured angular rate SS g can be expressed in reference coordinates by     0 0 (2) = SE q[k] ⊗ ⊗ SE q −1 [k]. Eg Sg The angle between the y-axis of the IMU and the vertical z-axis E y of the reference frame is equal to the trunk roll angle plus 90 ◦ /s. Hence, using only the z-component of S E y, we can determine the trunk roll angle by π (3) φ[k] = arccos(E q[k]z ) − . 2 Likewise, the time derivative of the trunk angle is obtained by projecting the angular rate E g onto the roll axis of the trunk, which is perpendicular to S y and E z:   E S Ez × Ey ˙ = g· , (4) φ[k] E � EE z�2 � SE y�2 According to Sanders et al. (2017) and Deschodt et al. (1999), the leg kick in front crawl swimming plays a major role for keeping the body in a streamline position. The type of synchronization depends on the skill level of the swimmer since expert swimmers execute several flutter kicks during one arm movement. For our subjects, we assume a slow swimming movement and synchronize the knee extension of each side to the forward movement of the contralateral arm. To realize the roll angle-triggered stimulation, we designed a state machine based on the sensor information (cf. Fig. 6) which starts a stimulation phase of 0.4 s as soon as the trunk roll angle exceeds 15 ◦ at an angular velocity of more than 40 ◦ /s. All thresholds and stimulation durations are so far only first guesses and need to be adjusted for paraplegic swimmers individually. 5. PRELIMINARY EXPERIMENTAL RESULTS With this state machine, we performed an experiment with a healthy subject on a padded platform where the trunk extends beyond the table to allow free shoulder movements. The IMU sensor was attached with a Velcro strap

IFAC CPHS 2018 282 Miami, FL, USA, Dec. 14-15, 2018

Constantin Wiesener et al. / IFAC PapersOnLine 51-34 (2019) 278–283

40 20 0 -20 -40 55

56

57

58

59

60

61

62

63

64

65

56

57

58

59

60

61

62

63

64

65

56

57

58

59

60

61

62

63

64

65

100 50 0 -50 -100 55 30

20

10

0 55

After further tests on healthy subjects, we plan to test the trigger method within the STIMSWIM study. STIMSWIM is a pilot study that aims at investigating the technical feasibility of FES to support swimming motions by the legs in up to 10 subjects with complete paralysis of the lower extremities after spinal trauma. There are three main questions which shall be answered in these trials. Does the swimming speed increase compared to non-assisted swimming? 2. Does the general well-being of the subject improve during the trial? 3. How is the acceptance of the technology by the user? Each trial comprises of a land training phase and a 12 session swimming phase. After the recruitment and initial assessment, the subject is asked to carry out a four-week FES cycling training at home. During this land training, he/she is asked to train at least three times a week for 30 minutes with a standard FES cycling ergometer (RehaMove, Hasomed GmbH, Germany). At the beginning and end of this training phase, the thigh diameter and maximum cycling power are assessed. This preliminary FES cycling training is needed to build up a defined baseline for the swimming trial. During the swimming phase, the FES cycling shall be reduced to two times a week. Furthermore, the subject was asked to fill in a questionnaire after each assessment. As a safety measure the swim sessions were always accompanied by a trained 338

pool guard. Furthermore, all recruited subjects are able to swim without stimulation. At the current state, two subjects completed the FES land training and the first swimming training and assessment sessions. Both subjects are ASIA impairment scale A with lesion level T5/6. The training was done at 16 m pool and one subject used a snorkel. In Fig. 8, the averaged results of the elapsed time for the swimming trials are shown. For both subjects, a clear training effect Elapsed time in seconds (16 m)

between the bladebones as shown in Fig. 1. For stimulation, the same stimulator as for the swimming experiment has been used, but only the left and right quadriceps were stimulated. In Fig. 7, the roll angle, velocity and the resulting stimulation currents of the experiment are presented. The subject was asked to perform a crawl movement with his arms and trunk. Whenever the stimulation criteria were fulfilled, the corresponding quadriceps muscle was stimulated.

Fig. 7. Experimental data (sampling frequency f = 100 Hz) for one healthy subject performing a crawl movement of the arms and trunk while the quadriceps muscles are stimulated (Stimulation frequency f = 25 Hz, pulse width of 300 μs).

Subject A

40

S1 without FES S1 with FES without TSCS S1 with FES with TSCS

35

30

25

20 0

10

20

30

40

50

60

70

Days after swimming trial start Elapsed time in seconds (16 m)

R Fig. 6. State machine implemented in Stateflow where entry defines a singular action when the state is entered and during defines a periodic action for each time index k, where cur left and cur right describe the scale of the output current for the left and right leg, dphi describes the roll angle velocity and after indicates in the Stateflow syntax that the state machine should leave the state after exact 0.4 s.

Subject B

40

S2 without FES S2 with FES without TSCS S2 with FES with TSCS

35

30

25

20 0

10

20

30

40

50

60

70

Days after swimming trial start

Fig. 8. Preliminary results of the first two subjects of the STIMSWIM study. can be observed. The difference between swimming speed with and without FES is ≈5 %. The difference between swimming speed with and without FES in combination with tSCS is ≈15 %. Additionally both subjects reported

IFAC CPHS 2018 Miami, FL, USA, Dec. 14-15, 2018

Constantin Wiesener et al. / IFAC PapersOnLine 51-34 (2019) 278–283

individually that swimming with tSCS reduced spasticity in the lower extremities for up to 4 hours. 6. DISCUSSION AND CONCLUSION A new concept for FES assisted-swimming in paraplegics was proposed which uses a waterproof stimulator and electrodes to produce a swimming movement of the paralyzed legs. During underwater experiments, a periodic FESinduced extension of the knee was achieved, which lead to a propulsion movement. To synchronize the leg movements with the upper body movement, an inertial sensor can be attached to the trunk, and the roll angle of the trunk can be used to trigger the stimulation in future. As trigger criterion, a combination of the absolute roll angle of the trunk and the angular velocity is proposed. So far we only tested the new method on a healthy subject, and we were able to trigger the stimulation of the contralateral side. This new method shall be tested with paraplegic subjects in the SWIMSTIM study in future. The ongoing study will include validation of the results in a larger number of paraplegic patients. If it is possible to show that an effective swimming training including the paralyzed legs can be realized, a completely new aqua therapy for paraplegics can be established. In addition, the stimulation could also be used recreationally by paraplegics for swimming and diving. Furthermore, it is conceivable that not only complete paraplegic patients but also incomplete paraplegic patients or stroke patients could benefit from an FES-assisted gait therapy in water. An improvement in physical functions and walking ability in manual underwater training has been shown in several studies (Tamburella et al., 2013; Stevens et al., 2015). 7. ACKNOWLEDGMENT We would like to acknowledge Axelgaard Manufacturing Co., Ltd., USA for developing, producing and donating the stimulation electrodes. Furthermore, we would like to thank all participants and partners in the STIMSWIM clinical trial for their support and commitment. REFERENCES Axelgaard, J. (2004). Current-Controlling Electrode with Adjustable Contact Area. US 6,745,082. Axelgaard, J. (2010). Moisture Resistant Electrode with Edge Protection. US 7,697,999. B¨achlin, M., F¨ orster, K., and Tr¨ oster, G. (2009). SwimMaster: A Wearable Assistant for Swimmer. In Proceedings of the 11th international conference on Ubiquitous computing - Ubicomp ’09, 215. ACM Press, New York, New York, USA. Bromley, I. (2006). Tetraplegia and paraplegia : a guide for physiotherapists. Churchill Livingstone. Brunelli, G. (2014). Topics in Paraplegia. fig 1. InTech, 1 edition. Callaway, A.J., Cobb, J.E., and Jones, I. (2009). A Comparison of Video and Accelerometer Based Approaches Applied to Performance Monitoring in Swimming. International Journal of Sports Science & Coaching, 4(1), 139–153. 339

283

Deschodt, J.V., Arsac, L.M., and Rouard, A.H. (1999). Relative contribution of arms and legs in humans to propulsion in 25-m sprint front-crawl swimming. European Journal of Applied Physiology and Occupational Physiology, 80(3), 192–199. Gibbons, R.S., Stock, C.G., Andrews, B.J., Gall, A., and Shave, R.E. (2016). The effect of FES-rowing training on cardiac structure and function: Pilot studies in people with spinal cord injury. Spinal Cord, 54(10), 822–829. Hofstoetter, U.S., McKay, W.B., Tansey, K.E., Mayr, W., Kern, H., and Minassian, K. (2014). Modification of spasticity by transcutaneous spinal cord stimulation in individuals with incomplete spinal cord injury. The Journal of Spinal Cord Medicine, 37(2), 202–211. Nash, M.S. (2005). Exercise as a health-promoting activity following spinal cord injury. Journal of Neurologic Physical Therapy, 29(2), 87–106. Newham, D.J. and Donaldson, N.D.N. (2007). FES cycling. Acta Neurochirurgica, Supplementum, 97(97 PART 1), 395–402. Psycharakis, S.G. and Sanders, R.H. (2010). Body roll in swimming: A review. Journal of Sports Sciences, 28(3), 229–236. Sanders, R.H., Andersen, J.T., and Takagi, H. (2017). The Segmental Movements in Front Crawl Swimming. In Handbook of Human Motion, 1–15. Springer International Publishing, Cham. Schauer, T. (2017). Sensing motion and muscle activity for feedback control of functional electrical stimulation: Ten years of experience in Berlin. Annual Reviews in Control, 44, 355–374. Seel, T. and Ruppin, S. (2017). Eliminating the Effect of Magnetic Disturbances on the Inclination Estimates of Inertial Sensors. IFAC-PapersOnLine, 50(1), 8798– 8803. Seifert, L., L’Hermette, M., Wattebled, L., Komar, J., Mell, F., Gomez, D., and Caritu, Y. (2011). Use of Inertial Central to Analyse Skill of Inter-Limb Coordination in Sport Activities. BIO Web of Conferences, 1, 1–4. Seifert, L., Leblanc, H., Chollet, D., and Deligni`eres, D. (2010). Inter-limb coordination in swimming: Effect of speed and skill level. Human Movement Science, 29(1), 103–113. Stevens, S.L., Caputo, J.L., Fuller, D.K., and Morgan, D.W. (2015). Effects of underwater treadmill training on leg strength, balance, and walking performance in adults with incomplete spinal cord injury. The Journal of Spinal Cord Medicine, 38(1), 91–101. Tamburella, F., Scivoletto, G., Cosentino, E., and Molinari, M. (2013). Walking in Water and on Land After an Incomplete Spinal Cord Injury. American Journal of Physical Medicine & Rehabilitation, 92(10 Suppl 2), e4–e15. Tawashy, A.E., Eng, J.J., Lin, K.H., Tang, P.F., and Hung, C. (2009). Physical activity is related to lower levels of pain, fatigue and depression in individuals with spinalcord injury: A correlational study. Spinal Cord, 47(4), 301–306. Wiesener, C. and Schauer, T. (2017). The Cybathlon RehaBike: Inertial-Sensor-Driven Functional Electrical Stimulation Cycling by Team Hasomed. IEEE Robotics & Automation Magazine, 24(4), 49–57.