Developing and Implementing Force Transducers in Gross Motor Tasks as a Means of Quantitative Assessment
Dr Edmund Biden
Dr Victoria Chester
Wendy Hill (BSc OT)
Force transducers have been created and are currently being used to investigate joint loading in children aged four to fourteen through the use of grossmotor tasks. Joint loading will be examined by combining data captured in by a motion capture system regarding the orientation of the child while bicycling or swinging and the data captured by the force transducers during the task.
The force transducers have been implemented on a swing and on a bicycle. On the swing, force transducers were located at hand-grip height along the chains, while on the bicycle, the hand grips were instrumented. The swing transducers provide grip and shear force (the amount of "down force" that is being applied to the chains), independently, on each side. On the bicycle, the transducers measure grip, force and torque independently, on each side. Fifteen normally limbed children and five children using a prosthesis have been tested; the prosthesis users have performed the protocol while wearing their prosthesis, and again without it. The results have shown that the force transducers are fully functional, and significant data can be collected through the tasks.
When examining positional and force data, certain trends for prosthesis users appear. These trends can be compared against the data obtained from the normal population. Since a small number of children using a prosthesis have been tested, each child will be treated as a case and compared individually against the normal database. As this number increases, it may become possible to compare populations.
Upper limb deficiency is a rare condition that requires lifelong efforts at adaptation. According to the Shriner's Hospitals for Children, congenital deficiencies affecting a single side below the elbow are the most common upper limb deficiency in North America (Davids, 2006). The deficiency may occur at the proximal, middle or distal portion of forearm. The New Brunswick Limb Deficiency Clinic is the referral centre for children and adults with upper limb deficiency and is run jointly by the University of New Brunswick's Institute of Biomedical Engineering (IBME) and the province's tertiary rehabilitation centre, the Stan Cassidy Centre for Rehabilitation (SSCR). Their statistics show that the majority (57%) of children seen with upper limb deficiencies were born missing the limb (Hill, 2008). The National Amputee Centre (War Amps) has similar statistics to those of the New Brunswick Limb Deficiency Clinic, reporting that 58% of their clients in the CHAMP program (which includes children under the age of eighteen) are upper limb amputees, with 60% due to congenital defects and the remaining 40% split evenly between being due to trauma or illness (Grierson, 2008). Annually, the New Brunswick Limb Deficiency Clinic enrolls 2-3 new children with an upper limb deficiency; in Canada, approximately 40 children are born each year with a congenital upper limb deficiency requiring prosthetic treatment (Ross, 2005).
In the case of an arm deficiency, daily tasks and activities can be improved through the use of a prosthesis, which replaces some functions of the absent limb. In Canada, most limb-deficient children are fitted early in life with a prosthesis, to aid with functioning independently in daily activities (Pruitt, 1999). Since a prosthesis has limited degrees of freedom, the use of an upper limb prosthesis often requires awkward postures in order to position the limb for functional tasks. Because of the compensatory motions necessary to accommodate these postures, the user's body may be subject to excessive loads, especially on their sound side. Therefore, it is important to consider the risks of prosthesis use.
There exists within the field of prosthetics a need for specific and accurate testing protocols and analysis methods. Few methods of quantitative testing currently exist, yet this sort of testing would provide information to assess the efficacy of new and possibly costly devices, as well as to assess the risks of prosthesis use. The information gained from quantitative tests tells not only how the device itself is functioning; it enables research into how to develop the technology to emulate the functions of a real limb, reduce the strain of a prosthesis on the human body, and understand the motion of the human body in general.
A protocol using quantitative tests has been developed to answer this need. After realizing that the traditional tasks used for ergonomic assessment held little appeal for children, gross motor tasks including swinging and riding a bike were created, and the equipment used for these tasks has been modified to include transducers, which allow the measurement of key kinematic and kinetic elements.
Several tasks have been developed through previous IBME Masters projects to examine gross motor movement. Experiments using these tasks not only measure positional data; through senior undergraduate projects, force transducers have been created that provide an upper limb force measurement analogous to force plates. These transducers are incorporated into a swinging task and a bike task.
The swinging task provides cyclic data that makes inter and intra-subject comparisons easier. A cycle begins with the subject leaning back and pumping his or her legs to swing forward, then leaning forward to swing back. Symmetry is very important in this task: if asymmetry occurs between the sides of the body, twisting may occur. This encourages the prosthesis-using subjects to emulate their normal side with their prosthesis using side. Timing is also important in this task in order to be able to pump to gain amplitude. As well, the swing has a fixed period determined by geometry. The swinging task is shown in Fig. 1 .
The bicycle task is slightly more involved than the swinging task. It examines stamina, balance and coordination, and is difficult to do without a prosthesis, in the case where the subject is a prosthesis user. The bicycle is quasi-stationary: spring-loaded struts on either side of the rear wheel require that the rider balance him/herself, and encourage the prosthesis to be actively used, not just rested upon the handlebars. Instructions are given throughout the task regarding the direction and speed, which test the rider's balance and coordination of body movements, range of motion and timing. The bicycle configuration is shown in Fig. 2 .
Both of these tasks are outfitted with transducer systems, which work in a similar fashion. For the bicycle, the transducers (shown in Fig. 3 ) work by having an outer shell with a gap cut into it which closes when the grip is held. Strain gauges measure the deflection within the inner wall, allowing the grip force to be calculated. A solid core, attached to a rectangular bar, measures the forward and downward forces using another set of strain gauges. The torsional transducers consist of a hollow aluminum piece positioned over a cut in the handlebars, with strain gauges placed to cancel out any bending or shear, measuring only torsion. Fig. 4 illustrates how the grip, forward and downward force transducers are designed.
Eight channels are used to collect data from the bicycle handgrip transducers: one each for grip, torsion, forward and downward forces for each hand. These data are taken into the VICON system, and then analyzed using Matlab.
The swing transducers function in a similar method to the bicycle transducers with a gap down one side of a cylindrical collar, as can be seen in Fig. 5 and Fig. 6 . When the gap closes, the deflection is measured using strain gauges located on the inner wall of the collar, which can be translated into grip force. The core of the transducer is rectangular and solid, attached to the rear wall of the collar with pins, which are outfitted with strain gauges, two on the top and two on the bottom. These strain gauges measure the strain on the pins as the collar is loaded relative to the core- that is, the amount of force with which the subject is pulling down. The transducers are placed in line with the chains, using shackles, to replace the section of chain at a height suitable for a child to hold.
Four channels are used to collect data from the swing transducers: one each for grip, forward downward forces for each hand.
These two tasks are incorporated into a series of tasks that are recorded using an eight camera Vicon 512 system. Six trials are captured for each task, and a marker set using 37 markers is used to capture the motion and show the movement of limb segments throughout each trial. The marker set is shown in Table 1 .
The procedure for the swinging tasks is quite simple–the height of the swing is set so that the child is able to place his or her feet on the ground, and then the child is instructed to swing normally. In the case of younger children, a push may be given at the beginning and slow down help may also be provided at the end, once all of the data have been collected for the trial. The children using a prosthesis perform this task twice, once while wearing their prosthesis, and once without it. This is to examine the symmetry and postural differences between the two cases.
The bicycle task has a slightly more involved procedure. Once the seat height is set to the correct level for the size of the subject, the child mounts the bicycle, and is given a series of instructions while cycling.
This set of instructions has variations every five seconds–first the handlebars are positioned as if cycling straight, then as if turning to the left, then back straight, then as if turning right, then straight again, and finally the child is asked to pedal while standing up. A marker on the knee shows the pedaling cycles, while markers on the hand show the turning segments of the trial. As with the swinging trial, children using a prosthesis perform this task twice, once while wearing their prosthesis, and once without it.
Results and Discussion
Before examining the results from the force transducers, it is interesting to examine the positional data to observe some of the postures taken by the prosthesis users during the trials. The norm (outside of the lab environment) for prosthesis users seems to be to wear the prosthesis while biking, but not while swinging. For one of the trials while biking without using the prosthesis, the subject was asked to touch the handlebars with his or her residual limb, as the tendency was to sit upright and control the handlebars only with the sound limb. The three figures in Fig. 7 show the biking task while wearing the prosthesis, without the prosthesis, and finally, without the prosthesis but touching the handlebars.
While the subject is wearing the prosthesis, the shoulders are close to being level, but not wearing the prosthesis results in a lean towards the sound limb, and not wearing the prosthesis but touching the handlebars results in a lean towards the residual limb.
During the swinging task, it was observed that there tended to be more twisting while using the prosthesis, as well as an awkward posture while holding the chains with a clear deviation toward the missing limb. Although somewhat difficult to see from the Vicon screen captures, Fig. 8 show swinging without a prosthesis, and swinging with a prosthesis (in which the beginning of some twisting behavior can be seen).
To contrast the previous images, Fig. 9 shows a normally limbed child performing the same tasks. Note that the body position is very symmetric and the shoulders are level.
The signal from the bicycle force transducer is segmented to distinguish the various moves the child is asked to make. This segmentation is based on the movement of the left hand ulna marker and the sacrum marker. The ulna marker clearly shows the straight/left/right segments of the bicycle trial, where the sacrum marker denotes the beginning of the standing segment by a drastic change in marker height. These lines are denoted in Fig. 10 . The knee marker is also plotted to show the pedaling cycles during normal pedaling ( Fig. 11 ), the plot looks sinusoidal, but interruptions to the regularity sometime occur, and tend to be the subject's foot slipping off the pedal, or a result of too sharp of a handlebar turn, which causes the knee to hit the handlebars, interrupting the regular cycles for a moment (shown in Fig. 12 ). The regularity can also be interrupted during the standing portion of the trial, as some subjects had trouble balancing on the semi-stationary bicycle (seen in Fig. 13 ). These three cases (normal and regular, regular but interrupted by a foot off the pedal, and irregular during standing phase) are shown in the three figures above.
The swing transducer data are segmented according to the extremes of the swing as a pendulum, that is, the high point when swinging forward is indicated, as well as the high point while swinging backward. This is easily seen in Fig. 14 . The lines marked at the bottom by a 'o' show when the subject is at his or her maximum in the forward direction, while the unmarked lines show when the subject is at his or her maximum amplitude in the backward portion of the swing.
The data from two representative cases (one normally limbed child and one prosthesis using child) will be presented in the following pages. In total, 15 normally limbed children and five prosthesis users have been tested. For demonstration purposes, only one from each group will be shown. The following figure shows the transducer data captured during a swinging trial with a normally limbed child. The five lines of plots represent, from the top down, the swinging cycles as represented by the movement of the knee marker, the squeeze force (measured in Newtons) exerted by the left hand, the left hand shear (downward pulling) force (measured in Newtons), followed by the squeeze force (measured in Newtons) exerted by the right hand, and the last line shows the right hand shear force (measured in Newtons).
It can be seen that the force magnitudes are similar between the two hands, with squeezing varying between 0 and 80 N, and shear varying between 0 and 65 N. This subject was right hand dominant. It is interesting to note that when looking at the Vicon position data, at the 60% trial mark, the maximum swinging amplitude is achieved. This coincides with the largest signals being produced.
The following figure shows the data captured during a prosthesis user swinging trial, while the subject was wearing her prosthesis (which is worn on the right side). The hand of the prosthesis was holding the force transducer, and the prosthesis was turned on.
The force magnitudes in this case are greater on the sound side, but the prosthesis is actively being used to hold on. When watching the data playback on the Vicon system, it can be seen that the subject was twisting substantially during the entire trial, which would be explained by the lack of symmetry between the two sides.
While not wearing her prosthesis, the data suggests that the subject was holding on more tightly with her sound side than she did while wearing her prosthesis. This can be seen in Fig. 15 . The Vicon data playback for this trial shows far less twisting for this case than the case where she was using her prosthesis. Also, it is important to note that the subject in this trial had a very low level amputation, with her residual limb ending at about the wrist, so there is a bit of force exerted by the residual limb.
Fig. 16 presents the data captured during the cycling trial with a normally limbed child. The first five lines of data represent the left side, and the subsequent five represent the right side. Within the five lines representative of each side, the first shows the ulnar marker movement, along with the turning/standing segment marks, as well as the pedaling cycles, represented by the movement of the knee marker. The second through fifth in each set show, from the top down, the squeeze force (measured in Newtons), the torque (twisting on the handgrips, measured in Newtons), how much the subject was pushing down on the handgrips (measured in Newtons), and how much the subject was pushing forward (measured in Newtons). The squeeze channel on the right side is somewhat noisy, but is functional. This will be improved in a future prototype.
The torsional transducer readings show very little torque exerted on the handgrips, ranging from zero to approximately 0.6 N, so it is basically zero. Similarities in shape and magnitude exist between the two sides, suggesting that this normally limbed subject is symmetric while cycling in this trial. The squeeze and downward push channels show an increased magnitude during the standing portion of the trial (approximately 85% to end). The prosthesis user shown by the data below is the same as was shown in the swing data. Fig. 17 shows the data for the trial performed while she was wearing her prosthesis.
The first five lines of Fig. 18 's data represent the left side (sound side), and the subsequent five represent the right side (prosthesis). Symmetry is present between the two sides while the prosthesis is being worn, and magnitude of force exerted is similar between the two sides since the prosthesis is actively being used. As had been seen previously with the normally limbed subject trial, the magnitude increases over all of the channels (except torsion, again it is very close to being zero) during the standing segment of the trial.
Fig. 19 presents the data captured while the subject was cycling without using her prosthesis. As the end of the residual limb was only resting on the handgrip, there is no squeeze force, but she was pushing down (X data) and forward (Y data) on the handgrip. Once again, very little torque is seen during the task. Surprisingly, there is symmetry between the push down and push forward forces on both sides, although squeeze is only present on her sound side. These forces are similar in magnitude to those seen while she was performing the task with her prosthesis. Again, the standing segment of the trial results in increased downward push force and squeeze force, similar to the magnitude of the forces seen while she was using her prosthesis.
Institute of Biomedical Engineering Fredericton, Bew Brunswick, Canada
- Davids, J. R. (2006). Prosthetic Management of Children with Unilateral Congenital Below-Elbow Deficiency. The Journal of Bone and Joint Surgery , 88-A (6), 1294-1300.
- Dewis, C. (2007). Measuring Upper Limb Loads During Activities of Daily Living. Fredericton: University of New Brunswick.
- Grierson, M. (2008, November 26). Personal Email.
- Hill, W. (2008, January 22). New Brunswick Limb Deficiency. Fredericton, NB: Personal Communication.
- Lake, C. a. (2006). Progressive Upper Limb Prosthetics. Physical Medicine and Rehabilitation Clinics of North America , 49-72.
- MacPhee, B. J. (2007). Examining the prosthetic function and body behavior of prosthesis users performing activities of daily living. Fredericton: University of New Brunswick. Dept. of Mechanical Engineering.
- National Instruments. (2007, June 7). Measuring Strain with Strain Gauges- Developer Zone. Retrieved November 2007, from National Instruments–Test and Measurement: http://zone.ni.com/devzone/cda/tut/p/id/3642
- Pruitt, S. D. (1999). Toddlers With Limb Deficiency: Conceptual Basis and Initial Application of a Functional Status Outcome Measure. Archives of Physical Medicine and Rehabilitation , 80, 819-824.
- Ross, M. (2005). Development of a Quantitative Test for Prosthetic Function Using Activities of Daily Living and Motion Analysis. Fredericton: University of New Brunswick