Kinematic Analysis of Unilateral and Bilateral Drinking Task after Brain and Periphery Injuries

Abstract

In this study, we aim to understand how functional movement is affected by various physical and neurological injuries. Using a novel system known as the Bilateral Assessment System (BiAS), we tracked the wrist movement of subjects with stroke, cerebral palsy, transradial amputations, as well as healthy individuals in space and in time as they completed a drinking task. The BiAS allowed us to analyze the kinematics of the movement of subjects to describe how specific impairment types at varying levels of the nervous system affect upper limb movement. Using kinematics to provide an objective analysis, the results suggest that bilateral ability is not analogous to unilateral ability and the importance of bilateral training, in addition to unilateral training, in the rehabilitation of injured persons is highlighted. In addition, motor, sensory, and strength impairment and subsequent ADL functional ability is a function of impairment level and not just impairment type.

I. Introduction

In 2011, 5.2 million adults required help with Activities of Daily Living (ADLs) such as eating or bathing and 9.8 million adults required help with instrumental activities of daily living (IADLs) such as household chores or shopping [1]. Brain and peripheral injuries are often the reason for these functional impairments in ADLs. For example, an ischemic stroke is a condition in which obstructed blood flow to the brain causes cell death to a portion of the brain and often results in weakened motor and/or sensory function of the affected side of the body and its limbs. Cerebral palsy (CP) refers to a set of congenital conditions resulting from an insult to the central nervous system leading to posture and motor impairment in addition to decreased muscle tone and spasticity [4,5]. Diabetes or traumatic injury often leads to limb loss in the form of an amputation; prosthetics serve to replace the lost portion of the limb with the aim of restoring some of the lost functional ability.

It is estimated that 6.8 million Americans above the age of 20 have had a stroke [2,3], about 800,000 children and adults have cerebral palsy [4,5] and about 350,000 have an amputation [6]. Brain and peripheral injuries will affect the ability of these individuals to control voluntary movement of the upper limb in different ways [7]. Important to this process is the ability to control the position of the upper extremity while reaching towards an object, the ability to control the orientation of the hand to grasp the object and finally the ability to control the fingers to grasp the object securely. Interlimb and intralimb muscle coordination, muscle strength, and sensory feedback are key to performing unilateral and bilateral tasks successfully.

The growing need to care for persons with functional impairments has led to developments in rehabilitation robotics. This field focuses on the development of devices that assist or assess persons who have suffered from neurological or physical injury that results in impaired motor abilities. The goal of technology-assisted therapy is often to help persons with brain injuries regain motor control over their impaired limb. Typically interventions focus on training the impaired limb only and assumes bilateral functional improvement as an outcome. Johnson and colleagues studied the wrist kinematics of post-stroke subjects in a bilateral drink and pour task after unilateral training with the ADLER robot therapy environment [8]. Their study provides evidence that for post-stroke patients, reduction in motor impairment due to unilateral training did not directly correlate to bilateral functional ability in bilateral drink and pour tasks. Coordination between the limbs, even if only one limb is impaired, may call for bilateral training either before, after or simultaneous to unilateral training. Waller and colleagues suggest that bilateral arm training is important for improved recovery, in addition to unilateral training [9]. However, a recent Cochrane review suggest that the evidence for bilateral training impact is not strong and that symmetric bilateral training had no effect on performance of ADLs; therefore, more research is needed [10].

In this study, we aim to examine unilateral versus bilateral performance in an ADL task by exploring how injuries such as stroke, CP, and amputations affect functional movement. By studying upper limb kinematics within and across these cohorts and comparing the results to that of healthy subjects, we gain insight into the influence of bilateral activity on the abilities of each hand, especially the impaired arm. We hypothesize that bilateral activity will have a positive effect on the activity of the impaired arm, however, subjects with greater motor impairment will need more time to complete tasks.

II. METHODS

A. Recruitment of Subjects

Twelve subjects participated in this study: five healthy subjects, four stroke subjects, two amputee subjects, and one CP subject. Table I details all subjects. Subjects were screened based on self-reports to ensure that they were at least 18 years old and that they had functional ability in both hands, such that they were able to complete the task. Furthermore, ischemic stroke survivors with a Fugl-Meyer score of no lower than 34 and amputee subjects with transradial amputation were included. Subjects gave informed consent. The Institutional Review Board at the University of Pennsylvania approved this study.

TABLE I.

Subject information

Subject	S 1	S 2	S 3	S 4	A 1	A 2	CP
Age	67	58	52	51	65	32	24
Dominant hand	RH	RH	RH	LH	LH	LH	LH
Box and Block D	44	52	51	28	20	62	20
Box and Block ND	53	30	18	27	59	27	11
Grip Strength D	65.1	53.8	81.6	51.5	64.5	94.2	52.4
Grip Strength ND	27.2	15.3	23.1	31.2	--	--	35.6
Mono filament D	4	4	4	8	10	2	7
Mono filament ND	5	7	5	11	35	35	4

B. Clinical Assessments

To quantify the functional ability of subjects, the subjects underwent clinical assessments. The Box and Block test (BBT), the grip strength (GS), and monofilament test can quantify hand dexterity (motor control), gross grip strength, and sensation, respectively, in all three populations [11–16]. The results of the clinical assessments serve to put all subjects, regardless of their injury, on to the same motor, sensory, and strength scales. The BBT is a test used by many occupational therapists, among others, to evaluate manual dexterity [17]. Strength was measured using the JAMAR^® Hydraulic Hand Dynamometer. The dynamometer is a tool that has been used to measure GS [18]. The Semmes-Weinstein monofilament was used to test sensory ability [19]. We used a hand set of 5 Baseline^® Tactile^™ Semmes-Weinstein Monofilaments that can apply a range of force of 2.83Nm to 6.65Nm.

C. Bilateral Assessment

The Bilateral Assessment System (BiAS) was originally developed to assess interlimb coordination after therapy. It is a low cost, portable, and user-friendly system that has the ability to dynamically assess wrist kinematics during ADLs [8] (Fig. 1).

Figure 1.

A screenshot of a subject seated at the workspace.

The BiAS robot consists of two 3 DOF mini-passive tracker robots, 1 for each wrist. As the robot trackers move, the raw values from the potentiometer are sent to a custom-made LabView program and post-processed in MATLAB to give the wrist position over time in Cartesian coordinates. The sampling rate was 100Hz. A custom-made button served as a timer. The subjects were instructed to press the button before and after each trial, sending a digital pulse of 5V to create distinctions between trials.

D. Drink Task

Subjects were seated in a chair at a distance from the table that enabled them to comfortably reach the back end of the table. Subjects were strapped into the chair to prevent trunk movement. The setup of the workspace is detailed in Fig. 1. Subjects performed a drink task, a commonly used ADL. Subjects were instructed to press the button, reach for the cup, bring it to their mouth, take a small sip of 150mL of water, return the cup to its original position and then press the button at home position. The cup was located on the subject midline at a distance of 13.5cm from the edge of the table. They were instructed to complete each of these tasks at a comfortable pace such as to ensure accuracy. The tasks started and ended with each hand on a button and was performed unilaterally and bilaterally, three times each in randomized order.

E. Data Analysis

A custom-made MATLAB program was used to perform the post-processing of the raw data acquired from the BiAS via LabView. The program applied a low-pass Chebychev filter to the raw values with a cutoff frequency of 10Hz. The actual start and end of a task was determined by the point in time in which the velocity dropped below 5% of the maximum velocity of the given trial. These cutoffs were applied to the filtered raw values. To analyze interlimb coordination and gain a deeper understanding of the effect of the impairment, algorithms were developed to calculate kinematic metrics [8]:

Task Completion Time (TCT): TCT is defined as the time in seconds that it takes to complete the task from initial button press to the final button press.

Phase Difference (PD): PD describes the lag between the two limbs as a measure of the synchronization of the two arms: dominant (less-impaired) phase less non-dominant (impaired) phase. Phase was calculated from the arctangent of the ratio of velocity to displacement. A positive PD signifies that the dominant arm was leading in the task.

Movement Overlap (MO): MO is defined as the time when both hands were in motion as a percentage of total TCT.

F. Statistical Analysis

Given our small sample size, for every calculated metric, a standard score was determined by dividing the difference between the average of the metric for the healthy subjects and the value of the subject in question by the standard deviation within the healthy population. Abnormal was characterized as having a standard score greater than or equal to 2 or less than or equal to −2, which is equivalent to greater than 2 standard deviations from the healthy subject mean.

III. Results

Fig. 2 shows a representative plot of the velocity profile of the non-dominant hand of 1 subject form each population during one of the trials. The task was split into four components, signifying the four phases in the drink task. Healthy subjects tended towards overall acceleration followed by overall deceleration for each phase and the peak velocity of each phase is similar.

Figure 2.

A representative plot of the velocity of the non-dominant hand during the first run of the unilateral trials: healthy 1, stroke 1, 1 cerebral palsy and 1 amputee subject.

For the stroke, cerebral palsy, and amputee subject, the maximum velocities in the initial and end phase is greater than that of the two middle phases. Namely, for the initial phase for the impaired subjects, especially the amputee subject, especially jerky, rapidly accelerating movement occurred. Bilateral Performance

Fig. 3 and 4 show the MO and PD of the impaired subjects as well as an average of the MO and PD of the healthy subjects. The error bars represent standard error.

Figure 3.

Bilateral Performance. Movement overlap (MO of all subjects during bilateral drink task. The average of all healthy is reported.

Figure 4.

Bilateral Performance. Phase difference of all subjects during bilateral drink task. The average of all healthy is reported

High movement overlap (MO) and low phase difference (PD) characterize good interlimb coordination for symmetric drink task. Healthy subjects had an MO of 67.43. Both amputee subjects and the CP subject performed abnormally with respect to MO (z-scores equal to 2.03, −3.42, 7.24, respectively). Healthy subjects had a positive PD of 2.46 whereas the PD of all impaired subjects were negative. Stroke subjects 3 and 4, Amputee 2 subject and the CP subject performed abnormally with respect to PD (z-scores equal to 4.33, 3.15, 6.36, 7.24, respectively). The healthy subjects clearly used their dominant hand to lead the task while impaired subjects led the task with their non-dominant hand.

Fig. 5 provides a further look at the effect of bilateral activity on the performance of the individual hands by looking at TCT of each hand during the unilateral and bilateral trials, respectively. The average of all healthy subjects is presented. The error bars represent standard error. The TCT of each hand in the bilateral trial was equivalent and is therefore represented by “Bi.” We expected that for the healthy subjects, all hands would perform similarly across both the unilateral and bilateral trials. In the impaired subjects, we expected that the dominant hand performing in a unilateral trial would perform best, such that it would have the lowest TCT. Furthermore, we expected that impaired subjects would perform worse than healthy subjects and that the dominant hand in the unilateral trial would perform best followed by the dominant hand in the bilateral trial. We expected the non-dominant hand in the unilateral trial would perform worst of all.

Figure 5.

Performance of Individual Arms. TCT of each hand (D meaning dominant hand and ND meaning non-dominant hand) during bilateral (Bi) and unilateral (Uni) activity for all subjects.

The expected increasing trend held true for all impaired subjects with the exception of Amputee 1 (TCT values of 5.52, 12.38, 8.30 sec for Uni-D, Bi, Bi-D, respectively) and the CP subject (TCT values of 15.68, 15.50, 12.29 sec for Uni-D, Bi, Bi-D, respectively). Further, while the performance of the non-dominant hand during the unilateral trial for all subjects was abnormal, the extent of abnormality decreased for all subjects except Amputee 1 (z-scores equal to −2.77 and −7.46 for Uni-ND and Bi-ND, respectively) and the CP subject in the bilateral trial (z-scores equal to −7.63 and −11.71 for Uni-ND and Bi-ND, respectively).

Fig. 6 relates motor control as quantified by the Box and Block (BBT) test to the performance of the non-dominant hand with respect to TCT. We expected that as the BBT score increased, TCT would decrease. The figure shows a line of best fit to the data and the corresponding correlation coefficient. R1 and its associated line of best fit includes the CP subject but R2 does not. Across all metrics, the fit was moderate. The trend showed that as BBT score increases, TCT decreases. While Stroke 3 supports the trend, the value is much greater than the fit would suggest (BBT score of 18 and a TCT of 28.38 sec) and the CP subject contradicts the trend (BBT score of 11 and a TCT of 15.50 sec).

Figure 6.

Motor Ability versus performance. TCT of the non-dominant hand during the unilateral trials compared to motor ability for all subjects. A linear fit was applied to the data and the correlation coefficient is reported (R). R1 represents the fit with the CP subject and R2 represents the fit without the CP subject.

IV. Discussion and conclusion

We aimed to confirm that interlimb coordination is affected post-brain and post-peripheral injury. We see that across the metrics, some or all subjects performed abnormally, providing support for this hypothesis. Overall, healthy subjects performed in a coordinated manner such that their movement was overlapped for a majority of the time but the dominant hand lead the task. Furthermore, each of their hands performed similarly, regardless of whether or not the task was unilateral or bilateral. While impaired subjects had substantial movement overall, the dominant hand did not lead and there was greater inconsistency as to how a limb would perform in a unilateral task versus a bilateral task. Overall, our data suggests that the non-dominant hand increases in performance while the dominant hand decreases in performance in the bilateral task. One could hypothesize that the dominant hand decreases its performance in an attempt to support the non-dominant hand, thereby enhancing the performance of the non-dominant hand [20].

Of particular interest in Fig. 6 is the observation of 4 clusters of subjects. These clusters formed with distinct separations and included subjects from different populations. Our data begin to suggest a lack of correlation between performance and injury type. For example, all stroke subjects did not have a consistent Box and Block score Lower functioning stroke subjects as determined by the BBT performed less normally than higher functioning subjects. A similar trend was observed in the amputees. This did not apply to CP. We begin to see linear correlations between an increasing Box and Block score and subjects’ performance, indifferent to injury type. Thus, our data suggest that functional performance as measured by kinematics and clinical evaluations might provide a universal platform for evaluating the effect of training with rehabilitation robotics across persons with brain and peripheral injuries [21]

The results from this study highlight the importance of bilateral training of the impaired arm to improve interlimb functional performance. It also highlights the importance of future robotic development to not only focus on the injury level, but also the manifestations of the injury. If a person requires strength training, a robotic device that provides this training is necessary, regardless of whether the patient is a stroke survivor, amputee, or cerebral palsy patient. In doing so, the robotic therapy increases utility. Next steps, will be to increase the sample size and evaluate more ADLs.